Why This Role Matters 

This role powers Tekion’s AI‑native, end‑to‑end automotive platform by turning unified dealership data across DMS, CRM, Digital Retail, Service, and Payments into real‑time intelligence. You’ll operationalize a graph‑based contextual ecosystem so agents can retrieve the right context, enforce policy, and personalize experiences that drive measurable dealer outcomes. You’ll also build the resilient control layer - MCP and the LLM Gateway - that enables safe, cost‑efficient, multi‑provider LLM usage. Finally, you’ll define the standards for building, evaluating, deploying, and governing agentic systems so product teams can ship AI features quickly, safely, and at scale. In addition to enabling agentic systems powered by LLMs, this role also drives building the platform for classical ML models driving optimization across dealership operations.  

What Makes This Opportunity Unique 

This role offers direct, measurable impact on dealer outcomes and consumer experiences across Tekion’s Automotive Retail Cloud and Automotive Enterprise Cloud, with end‑to‑end ownership of an LLM control plane and gateway that serve multi‑tenant workloads under SLAs and, quality and cost guardrails. You’ll leverage a rich vertical dataset and domain graph spanning sales, service, parts, F&I, accounting, and consumer touchpoints to power context‑aware agents and retrieval‑augmented generation. You’ll also shape core levers - agent orchestration patterns, evaluation frameworks, and safety guardrails - so improvements in latency, reliability, evaluation quality, and safety translate into dealer KPIs like upsell, cycle time, CSAT, and service revenue.  You’ll also maintain and enhance the platform to support classical supervised and unsupervised ML models . 

Responsibilities 

• Build and run the LLM control plane/gateway: smart routing, rate limits/quotas, failover, and token/cost tracking. 

• Ship a unified API and SDKs (REST/gRPC) with normalized schemas, structured outputs, caching, and full observability (traces/logs/metrics). 

• Enforce safety and privacy by default: content filtering, prompt/response validation, and PII redaction. 

• Enable multi‑model, multi‑vendor use LLMs with automated canarying and versioning. 

• Own the agent runtime: tool registry, permissions, function calling, grounding, and retrieval. 

• Design orchestration patterns (sequential, planner‑executor, streaming) and manage agent state and long‑running workflows. 

• Enabling platform components for training and scoring pipelines for classical ML (e.g., XGBoost/LightGBM/linear/trees) and deep models; standardize experiment tracking and packaging. 

• Create components to Monitor model and data drift, retraining and tuning models as needed to maintain accuracy and relevance. 

• Add human‑in‑the‑loop review and safe‑actioning before agents touch dealer systems. 

• Evolve the domain graph and entity resolution; build reliable data ingestion pipelines. 

• Serve real‑time context to agents (profiles, inventory, pricing, appointments, service history) with access controls and lineage. 

• Power retrieval with hybrid search (graph + vector + keyword) and smart cache/TTL to balance accuracy, latency, and cost. 

• Run continuous offline/online evaluations for quality, factuality, bias, and safety for the platform sanity. 

• Define SLOs for latency (p50/p95), uptime, and cost view capabilities; enable autoscaling and spend controls. 

• Maintain a model/agent registry, versioning, approvals, audit trails, and reproducibility; support compliances where needed. 

• Provide templates/CLIs, sandboxes, and docs so product teams can build and ship fast; mentor engineers and champion MLOps and AI safety best practices. 

Desired Skills & Experience 

• 12–15+ years building large‑scale data/ML or platform systems; strong software engineering fundamentals (Abstracted API design, concurrency, distributed systems). 

• Production experience with Python plus one of Java/Scala/Go; microservices and API design. 

• MLOps at scale: pipelines (Airflow/Kubeflow), tracking/registry (MLflow), CI/CD for models, A/B testing, shadow/canary, and online feature computation (Spark/Flink/Kafka). 

• Cloud and containers: AWS (preferred), plus Docker/Kubernetes; performance, reliability, and cost engineering in multi‑tenant SaaS. 

• Practical ML knowledge (feature engineering, training, evaluation, drift detection); experience deploying models that power user‑facing workflows. 

• Built or operated an LLM gateway/control plane: provider adapters, routing/policies, caching, quota/rate‑limit, cost and token accounting. 

• Agentic systems: tool use/function calling, orchestration frameworks, human‑in‑the‑loop, safety/guardrails, and online evaluation/telemetry. 

• Graph and retrieval: knowledge graphs (e.g., Neo4j/Neptune/TigerGraph), GraphQL, vector search (e.g., pgvector/Qdrant/Milvus), hybrid retrieval patterns. 

Preferred Mindset 

• Platform‑as‑product: obsess over developer experience, paved roads, and clear SLAs. 

• Thinks in systems - observability, fallback, access control are core, not afterthoughts. 

• Passionate about AI - enjoys enabling real-world LLM and agentic use cases. 

• Cost‑aware builder: you treat latency and dollars as first‑class metrics and design for graceful degradation. 

• Vendor‑agnostic thinker: choose the right model/provider per use case; build for portability and resilience. 

• Documentation and teaching: you make complex systems understandable; you uplevel teams. 

Job Title: Knowledge Graph

About Us

“Capco, a Wipro company, is a global technology and management consulting firm. Awarded with Consultancy of the year in the British Bank Award and has been ranked Top 100 Best Companies for Women in India 2022 by Avtar & Seramount. With our presence across 32 cities across globe, we support 100+ clients across banking, financial and Energy sectors. We are recognized for our deep transformation execution and delivery. 

WHY JOIN CAPCO?

You will work on engaging projects with the largest international and local banks, insurance companies, payment service providers and other key players in the industry. The projects that will transform the financial services industry.

MAKE AN IMPACT

Innovative thinking, delivery excellence and thought leadership to help our clients transform their business. Together with our clients and industry partners, we deliver disruptive work that is changing energy and financial services.

#BEYOURSELFATWORK

Capco has a tolerant, open culture that values diversity, inclusivity, and creativity.

CAREER ADVANCEMENT

With no forced hierarchy at Capco, everyone has the opportunity to grow as we grow, taking their career into their own hands.

DIVERSITY & INCLUSION

We believe that diversity of people and perspective gives us a competitive advantage.

Location - Bangalore

Experience - 6 to 10

Skills and Qualifications:

Preferred skills:

Role: Knowledge graphs

Mandatory skills: Knowledge-graph, RDF, Sparql, Neo4j

Staff Data Scientist – Supply Chain AI (Small Language Models)  

At Avathon, we build AI products that solve real industrial problems at scale. We believe the future of enterprise AI lies in small, focused, high-performance models—deeply grounded in domain context—not just large, generic systems. 

We’re looking for an experienced Data Scientist to join our Supply Chain AI team in India. In this role, you’ll design, fine-tune, and deploy optimized Small Language Models (SLMs) that power production-grade AI features across logistics, planning, inventory, and demand forecasting. 

What Makes This Role Unique 

This is a product-first data science role. You won’t just build models—you’ll own how they show up in real products used by large enterprises. Your work becomes the intelligence layer that sits on top of our graph and indexing systems and directly drives customer outcomes. 

What You’ll Do 

• Build and fine-tune SLMs/LLMs for production use cases such as semantic search, forecasting, contract intelligence, and conversational insights. Optimize models for performance, cost, and latency using PEFT, quantization, and efficient inference techniques 

• Embed supply chain knowledge into models by working closely with domain experts and product teams 

• Enable agentic workflows, allowing AI Assistants to execute planning, optimization, and decision-support tasks 

• Integrate models into products using custom Lambdas, and a Graph-based microservices architecture  

• Apply vector search and semantic reasoning to help customers navigate complex supply chain relationships 

• Work with real enterprise data—ERP, logistics, inventory, supplier, and contract datasets 

• Deploy and operate models using strong MLOps practices on platforms like AWS, GCP or AZURE 

• Measure and improve quality, reducing hallucinations and improving reliability through continuous evaluation 
• Partner with engineering, product, and operations teams to ship impactful AI features end-to-end 

What We’re Looking For 

• Experience working with SLMs (Mistral, Phi, Nemotron, Llama variants), applying parameter-efficient fine-tuning (LoRA/QLoRA) to optimize task-specific natural language query answering with high accuracy and minimized hallucinations. 

• Strong understanding of supply chain data and workflows (logistics, inventory, procurement, ERP systems) 

• Familiarity with Knowledge Graphs and graph-based data modeling, including designing and querying Neo4j graphs for relationship-driven insights. 

• Solid hands-on skills in Python, PyTorch/TensorFlow, GraphQL, and vector databases 

• 8+ years of experience in forecasting, optimization, supply chain analytics, or AI-driven products 

• Experience designing Retrieval-Augmented Generation pipelines using document chunking, metadata filtering, hybrid (dense + sparse) retrieval, and SLM/LLM grounding for enterprise knowledge search. 

• Hands-on experience building and tuning FAISS-based vector indexes (IVF, HNSW, PQ) for large-scale semantic similarity search with low-latency production deployment 

• Experience taking models from prototype to production 

• Bonus: reinforcement learning, MARL, graph neural networks, or probabilistic modeling 

• Clear communication skills and a strong sense of product ownership 

Why Avathon 

You’ll work on core AI products, not experiments—solving complex supply chain problems for global enterprises, while building from India as a key innovation hub. 

If you enjoy shipping AI that delivers measurable business value, this role is for you. 

Job Title: AI Full stack developer

 About Us

“Capco, a Wipro company, is a global technology and management consulting firm. Awarded with Consultancy of the year in the  British Bank Award and has been ranked Top 100 Best Companies for Women in India 2022 by Avtar & Seramount. With our presence across 32 cities across globe, we support 100+ clients across banking, financial and Energy sectors. We are recognized for our deep transformation execution and delivery. 

WHY JOIN CAPCO? 

You will work on engaging projects with the largest international and local banks, insurance companies, payment service providers and other key players in the industry. The projects that will transform the financial services industry.

MAKE AN IMPACT

Innovative thinking, delivery excellence and thought leadership to help our clients transform their business. Together with our clients and industry partners, we deliver disruptive work that is changing energy and financial services.

#BEYOURSELFATWORK

Capco has a tolerant, open culture that values diversity, inclusivity, and creativity.

CAREER ADVANCEMENT

With no forced hierarchy at Capco, everyone has the opportunity to grow as we grow, taking their career into their own hands.

DIVERSITY & INCLUSION

We believe that diversity of people and perspective gives us a competitive advantage.

Job Description:

Senior Software Engineer  -  Senior Machine Learning Platform

Why This Role Matters 

This role powers Tekion’s AI‑native, end‑to‑end automotive platform by turning unified dealership data across DMS, CRM, Digital Retail, Service, and Payments into real‑time intelligence. You’ll operationalize a graph‑based contextual ecosystem so agents can retrieve the right context, enforce policy, and personalize experiences that drive measurable dealer outcomes. You’ll also build the resilient control layer - MCP and the LLM Gateway - that enables safe, cost‑efficient, multi‑provider LLM usage. Finally, you’ll define the standards for building, evaluating, deploying, and governing agentic systems so product teams can ship AI features quickly, safely, and at scale. In addition to enabling agentic systems powered by LLMs, this role also drives building the platform for classical ML models driving optimization across dealership operations.  

What Makes This Opportunity Unique 

This role offers direct, measurable impact on dealer outcomes and consumer experiences across Tekion’s Automotive Retail Cloud and Automotive Enterprise Cloud, with end‑to‑end ownership of an LLM control plane and gateway that serve multi‑tenant workloads under SLAs and, quality and cost guardrails. You’ll leverage a rich vertical dataset and domain graph spanning sales, service, parts, F&I, accounting, and consumer touchpoints to power context‑aware agents and retrieval‑augmented generation. You’ll also shape core levers - agent orchestration patterns, evaluation frameworks, and safety guardrails - so improvements in latency, reliability, evaluation quality, and safety translate into dealer KPIs like upsell, cycle time, CSAT, and service revenue.  You’ll also maintain and enhance the platform to support classical supervised and unsupervised ML models . 

Responsibilities 

• Build and run the LLM control plane/gateway: smart routing, rate limits/quotas, failover, and token/cost tracking. 

• Ship a unified API and SDKs (REST/gRPC) with normalized schemas, structured outputs, caching, and full observability (traces/logs/metrics). 

• Enforce safety and privacy by default: content filtering, prompt/response validation, and PII redaction. 

• Enable multi‑model, multi‑vendor use LLMs with automated canarying and versioning. 

• Own the agent runtime: tool registry, permissions, function calling, grounding, and retrieval. 

• Design orchestration patterns (sequential, planner‑executor, streaming) and manage agent state and long‑running workflows. 

• Enabling platform components for training and scoring pipelines for classical ML (e.g., XGBoost/LightGBM/linear/trees) and deep models; standardize experiment tracking and packaging. 

• Create components to Monitor model and data drift, retraining and tuning models as needed to maintain accuracy and relevance. 

• Add human‑in‑the‑loop review and safe‑actioning before agents touch dealer systems. 

• Evolve the domain graph and entity resolution; build reliable data ingestion pipelines. 

• Serve real‑time context to agents (profiles, inventory, pricing, appointments, service history) with access controls and lineage. 

• Power retrieval with hybrid search (graph + vector + keyword) and smart cache/TTL to balance accuracy, latency, and cost. 

• Run continuous offline/online evaluations for quality, factuality, bias, and safety for the platform sanity. 

• Define SLOs for latency (p50/p95), uptime, and cost view capabilities; enable autoscaling and spend controls. 

• Maintain a model/agent registry, versioning, approvals, audit trails, and reproducibility; support compliances where needed. 

• Provide templates/CLIs, sandboxes, and docs so product teams can build and ship fast; mentor engineers and champion MLOps and AI safety best practices. 

Desired Skills & Experience 

• 5 - 7 years building large‑scale data/ML or platform systems; strong software engineering fundamentals (Abstracted API design, concurrency, distributed systems). 

• Production experience with Python plus one of Java/Scala/Go; microservices and API design. 

• MLOps at scale: pipelines (Airflow/Kubeflow), tracking/registry (MLflow), CI/CD for models, A/B testing, shadow/canary, and online feature computation (Spark/Flink/Kafka). 

• Cloud and containers: AWS (preferred), plus Docker/Kubernetes; performance, reliability, and cost engineering in multi‑tenant SaaS. 

• Practical ML knowledge (feature engineering, training, evaluation, drift detection); experience deploying models that power user‑facing workflows. 

• Built or operated an LLM gateway/control plane: provider adapters, routing/policies, caching, quota/rate‑limit, cost and token accounting. 

• Agentic systems: tool use/function calling, orchestration frameworks, human‑in‑the‑loop, safety/guardrails, and online evaluation/telemetry. 

• Graph and retrieval: knowledge graphs (e.g., Neo4j/Neptune/TigerGraph), GraphQL, vector search (e.g., pgvector/Qdrant/Milvus), hybrid retrieval patterns. 

Preferred Mindset 

• Platform‑as‑product: obsess over developer experience, paved roads, and clear SLAs. 

• Thinks in systems - observability, fallback, access control are core, not afterthoughts. 

• Passionate about AI - enjoys enabling real-world LLM and agentic use cases. 

• Cost‑aware builder: you treat latency and dollars as first‑class metrics and design for graceful degradation. 

• Vendor‑agnostic thinker: choose the right model/provider per use case; build for portability and resilience. 

• Documentation and teaching: you make complex systems understandable; you uplevel teams. 

Why This Role Matters 

This role powers Tekion’s AI‑native, end‑to‑end automotive platform by turning unified dealership data across DMS, CRM, Digital Retail, Service, and Payments into real‑time intelligence. You’ll operationalize a graph‑based contextual ecosystem so agents can retrieve the right context, enforce policy, and personalize experiences that drive measurable dealer outcomes. You’ll also build the resilient control layer - MCP and the LLM Gateway - that enables safe, cost‑efficient, multi‑provider LLM usage. Finally, you’ll define the standards for building, evaluating, deploying, and governing agentic systems so product teams can ship AI features quickly, safely, and at scale. In addition to enabling agentic systems powered by LLMs, this role also drives building the platform for classical ML models driving optimization across dealership operations.  

What Makes This Opportunity Unique 

This role offers direct, measurable impact on dealer outcomes and consumer experiences across Tekion’s Automotive Retail Cloud and Automotive Enterprise Cloud, with end‑to‑end ownership of an LLM control plane and gateway that serve multi‑tenant workloads under SLAs and, quality and cost guardrails. You’ll leverage a rich vertical dataset and domain graph spanning sales, service, parts, F&I, accounting, and consumer touchpoints to power context‑aware agents and retrieval‑augmented generation. You’ll also shape core levers - agent orchestration patterns, evaluation frameworks, and safety guardrails - so improvements in latency, reliability, evaluation quality, and safety translate into dealer KPIs like upsell, cycle time, CSAT, and service revenue.  You’ll also maintain and enhance the platform to support classical supervised and unsupervised ML models . 

Responsibilities 

• Build and run the LLM control plane/gateway: smart routing, rate limits/quotas, failover, and token/cost tracking. 

• Ship a unified API and SDKs (REST/gRPC) with normalized schemas, structured outputs, caching, and full observability (traces/logs/metrics). 

• Enforce safety and privacy by default: content filtering, prompt/response validation, and PII redaction. 

• Enable multi‑model, multi‑vendor use LLMs with automated canarying and versioning. 

• Own the agent runtime: tool registry, permissions, function calling, grounding, and retrieval. 

• Design orchestration patterns (sequential, planner‑executor, streaming) and manage agent state and long‑running workflows. 

• Enabling platform components for training and scoring pipelines for classical ML (e.g., XGBoost/LightGBM/linear/trees) and deep models; standardize experiment tracking and packaging. 

• Create components to Monitor model and data drift, retraining and tuning models as needed to maintain accuracy and relevance. 

• Add human‑in‑the‑loop review and safe‑actioning before agents touch dealer systems. 

• Evolve the domain graph and entity resolution; build reliable data ingestion pipelines. 

• Serve real‑time context to agents (profiles, inventory, pricing, appointments, service history) with access controls and lineage. 

• Power retrieval with hybrid search (graph + vector + keyword) and smart cache/TTL to balance accuracy, latency, and cost. 

• Run continuous offline/online evaluations for quality, factuality, bias, and safety for the platform sanity. 

• Define SLOs for latency (p50/p95), uptime, and cost view capabilities; enable autoscaling and spend controls. 

• Maintain a model/agent registry, versioning, approvals, audit trails, and reproducibility; support compliances where needed. 

Desired Skills & Experience 

• 4 - 5+ years building large‑scale data/ML or platform systems; strong software engineering fundamentals (Abstracted API design, concurrency, distributed systems). 

• Production experience with Python plus one of Java/Scala/Go; microservices and API design. 

• MLOps at scale: pipelines (Airflow/Kubeflow), tracking/registry (MLflow), CI/CD for models, A/B testing, shadow/canary, and online feature computation (Spark/Flink/Kafka). 

• Cloud and containers: AWS (preferred), plus Docker/Kubernetes; performance, reliability, and cost engineering in multi‑tenant SaaS. 

• Practical ML knowledge (feature engineering, training, evaluation, drift detection); experience deploying models that power user‑facing workflows. 

• Built or operated an LLM gateway/control plane: provider adapters, routing/policies, caching, quota/rate‑limit, cost and token accounting. 

• Agentic systems: tool use/function calling, orchestration frameworks, human‑in‑the‑loop, safety/guardrails, and online evaluation/telemetry. 

• Graph and retrieval: knowledge graphs (e.g., Neo4j/Neptune/TigerGraph), GraphQL, vector search (e.g., pgvector/Qdrant/Milvus), hybrid retrieval patterns. 

Preferred Mindset 

• Platform‑as‑product: obsess over developer experience, paved roads, and clear SLAs. 

• Thinks in systems - observability, fallback, access control are core, not afterthoughts. 

• Passionate about AI - enjoys enabling real-world LLM and agentic use cases. 

• Cost‑aware builder: you treat latency and dollars as first‑class metrics and design for graceful degradation. 

• Vendor‑agnostic thinker: choose the right model/provider per use case; build for portability and resilience. 

Why This Role Matters 

This role powers Tekion’s AI‑native, end‑to‑end automotive platform by turning unified dealership data across DMS, CRM, Digital Retail, Service, and Payments into real‑time intelligence. You’ll operationalize a graph‑based contextual ecosystem so agents can retrieve the right context, enforce policy, and personalize experiences that drive measurable dealer outcomes. You’ll also build the resilient control layer - MCP and the LLM Gateway - that enables safe, cost‑efficient, multi‑provider LLM usage. Finally, you’ll define the standards for building, evaluating, deploying, and governing agentic systems so product teams can ship AI features quickly, safely, and at scale. In addition to enabling agentic systems powered by LLMs, this role also drives building the platform for classical ML models driving optimization across dealership operations.  

What Makes This Opportunity Unique 

This role offers direct, measurable impact on dealer outcomes and consumer experiences across Tekion’s Automotive Retail Cloud and Automotive Enterprise Cloud, with end‑to‑end ownership of an LLM control plane and gateway that serve multi‑tenant workloads under SLAs and, quality and cost guardrails. You’ll leverage a rich vertical dataset and domain graph spanning sales, service, parts, F&I, accounting, and consumer touchpoints to power context‑aware agents and retrieval‑augmented generation. You’ll also shape core levers - agent orchestration patterns, evaluation frameworks, and safety guardrails - so improvements in latency, reliability, evaluation quality, and safety translate into dealer KPIs like upsell, cycle time, CSAT, and service revenue.  You’ll also maintain and enhance the platform to support classical supervised and unsupervised ML models . 

Responsibilities 

• Build and run the LLM control plane/gateway: smart routing, rate limits/quotas, failover, and token/cost tracking. 

• Ship a unified API and SDKs (REST/gRPC) with normalized schemas, structured outputs, caching, and full observability (traces/logs/metrics). 

• Enforce safety and privacy by default: content filtering, prompt/response validation, and PII redaction. 

• Enable multi‑model, multi‑vendor use LLMs with automated canarying and versioning. 

• Own the agent runtime: tool registry, permissions, function calling, grounding, and retrieval. 

• Design orchestration patterns (sequential, planner‑executor, streaming) and manage agent state and long‑running workflows. 

• Enabling platform components for training and scoring pipelines for classical ML (e.g., XGBoost/LightGBM/linear/trees) and deep models; standardize experiment tracking and packaging. 

• Create components to Monitor model and data drift, retraining and tuning models as needed to maintain accuracy and relevance. 

• Add human‑in‑the‑loop review and safe‑actioning before agents touch dealer systems. 

• Evolve the domain graph and entity resolution; build reliable data ingestion pipelines. 

• Serve real‑time context to agents (profiles, inventory, pricing, appointments, service history) with access controls and lineage. 

• Power retrieval with hybrid search (graph + vector + keyword) and smart cache/TTL to balance accuracy, latency, and cost. 

• Run continuous offline/online evaluations for quality, factuality, bias, and safety for the platform sanity. 

• Define SLOs for latency (p50/p95), uptime, and cost view capabilities; enable autoscaling and spend controls. 

• Maintain a model/agent registry, versioning, approvals, audit trails, and reproducibility; support compliances where needed. 

• Provide templates/CLIs, sandboxes, and docs so product teams can build and ship fast; mentor engineers and champion MLOps and AI safety best practices. 

Desired Skills & Experience 

• 5+ - 8+ years building large‑scale data/ML or platform systems; strong software engineering fundamentals (Abstracted API design, concurrency, distributed systems). 

• Production experience with Python plus one of Java/Scala/Go; microservices and API design. 

• MLOps at scale: pipelines (Airflow/Kubeflow), tracking/registry (MLflow), CI/CD for models, A/B testing, shadow/canary, and online feature computation (Spark/Flink/Kafka). 

• Cloud and containers: AWS (preferred), plus Docker/Kubernetes; performance, reliability, and cost engineering in multi‑tenant SaaS. 

• Practical ML knowledge (feature engineering, training, evaluation, drift detection); experience deploying models that power user‑facing workflows. 

• Built or operated an LLM gateway/control plane: provider adapters, routing/policies, caching, quota/rate‑limit, cost and token accounting. 

• Agentic systems: tool use/function calling, orchestration frameworks, human‑in‑the‑loop, safety/guardrails, and online evaluation/telemetry. 

• Graph and retrieval: knowledge graphs (e.g., Neo4j/Neptune/TigerGraph), GraphQL, vector search (e.g., pgvector/Qdrant/Milvus), hybrid retrieval patterns. 

Preferred Mindset 

• Platform‑as‑product: obsess over developer experience, paved roads, and clear SLAs. 

• Thinks in systems - observability, fallback, access control are core, not afterthoughts. 

• Passionate about AI - enjoys enabling real-world LLM and agentic use cases. 

• Cost‑aware builder: you treat latency and dollars as first‑class metrics and design for graceful degradation. 

• Vendor‑agnostic thinker: choose the right model/provider per use case; build for portability and resilience. 

Staff Data Scientist – Supply Chain AI (Small Language Models)  

At Avathon, we build AI products that solve real industrial problems at scale. We believe the future of enterprise AI lies in small, focused, high-performance models—deeply grounded in domain context—not just large, generic systems. 

We’re looking for an experienced Data Scientist to join our Supply Chain AI team in India. In this role, you’ll design, fine-tune, and deploy optimized Small Language Models (SLMs) that power production-grade AI features across logistics, planning, inventory, and demand forecasting. 

What Makes This Role Unique 

This is a product-first data science role. You won’t just build models—you’ll own how they show up in real products used by large enterprises. Your work becomes the intelligence layer that sits on top of our graph and indexing systems and directly drives customer outcomes. 

What You’ll Do 

• Build and fine-tune SLMs/LLMs for production use cases such as semantic search, forecasting, contract intelligence, and conversational insights. Optimize models for performance, cost, and latency using PEFT, quantization, and efficient inference techniques 

• Embed supply chain knowledge into models by working closely with domain experts and product teams 

• Enable agentic workflows, allowing AI Assistants to execute planning, optimization, and decision-support tasks 

• Integrate models into products using custom Lambdas, and a Graph-based microservices architecture  

• Apply vector search and semantic reasoning to help customers navigate complex supply chain relationships 

• Work with real enterprise data—ERP, logistics, inventory, supplier, and contract datasets 

• Deploy and operate models using strong MLOps practices on platforms like AWS, GCP or AZURE 

• Measure and improve quality, reducing hallucinations and improving reliability through continuous evaluation 
• Partner with engineering, product, and operations teams to ship impactful AI features end-to-end 

What We’re Looking For 

• Experience working with SLMs (Mistral, Phi, Nemotron, Llama variants), applying parameter-efficient fine-tuning (LoRA/QLoRA) to optimize task-specific natural language query answering with high accuracy and minimized hallucinations. 

• Strong understanding of supply chain data and workflows (logistics, inventory, procurement, ERP systems) 

• Familiarity with Knowledge Graphs and graph-based data modeling, including designing and querying Neo4j graphs for relationship-driven insights. 

• Solid hands-on skills in Python, PyTorch/TensorFlow, GraphQL, and vector databases 

• 8+ years of experience in forecasting, optimization, supply chain analytics, or AI-driven products 

• Experience designing Retrieval-Augmented Generation pipelines using document chunking, metadata filtering, hybrid (dense + sparse) retrieval, and SLM/LLM grounding for enterprise knowledge search. 

• Hands-on experience building and tuning FAISS-based vector indexes (IVF, HNSW, PQ) for large-scale semantic similarity search with low-latency production deployment 

• Experience taking models from prototype to production 

• Bonus: reinforcement learning, MARL, graph neural networks, or probabilistic modeling 

• Clear communication skills and a strong sense of product ownership 

Why Avathon 

You’ll work on core AI products, not experiments—solving complex supply chain problems for global enterprises, while building from India as a key innovation hub. 

If you enjoy shipping AI that delivers measurable business value, this role is for you.