Intellectsoft is a software development company delivering innovative solutions since 2007. We operate across North America, Latin America, the Nordic region, the UK, and Europe. We specialize in industries like Fintech, Healthcare, EdTech, Construction, Hospitality, and more, partnering with startups, mid-sized businesses, and Fortune 500 companies to drive innovation and scalability. Our clients include Jaguar Motors, Universal Pictures, Harley-Davidson, and many more where our teams are making daily impactTogether, our team delivers solutions that make a difference. Learn more at www.intellectsoft.net

Our client is a management consulting and technology firm focused on transforming global healthcare and diverse industries through data-driven strategies, cutting-edge analytics, and advanced AI solutions. This role sits within a focused research initiative dedicated to adapting small language models (SLMs) to specialized, high-value domain applications.

Requirements

• Experience: 3+ years in Machine Learning Engineering or Applied ML, with hands-on experience fine-tuning transformer-based language models.
• Core Tech Stack: Advanced Python and PyTorch proficiency.
• Frameworks & Ecosystem: Deep fluency in the Hugging Face ecosystem, including Transformers, PEFT, Datasets, Accelerate, and TRL.
• Fine-Tuning Expertise: Practical command of SFT with LoRA/QLoRA, alongside the judgment to reason about CPT and RL/preference methods.
• Model Selection: Strong understanding of base model architectures, tokenizer design, pre-training data distributions, licensing, and context window requirements.
• Data Engineering for ML: Proven experience building training pipelines (curation, cleaning, synthetic data generation, and formatting).
• Evaluation & Serving: Ability to design executable correctness evaluations, paired with hands-on experience serving and quantizing models using at least one modern inference framework.
• Research Fluency: Comfort reading recent ML research papers and translating state-of-the-art methods into working code.

Nice to have

• Hands-on experience with Continued Pre-Training (CPT) on large domain corpora.
• Experience with Reinforcement Learning / preference optimization for LLMs (reward modeling, DPO, GRPO, or verifiable-reward setups).
• Experience with structured-output generation, code generation (e.g., SQL/DDL), or domain classification fine-tuning.
• Familiarity with distributed training frameworks (DeepSpeed, FSDP).
• Knowledge distillation experience transferring capabilities from large teacher models to small student models.
• Experience with on-device or edge deployment.
• Active open-source contributions to machine learning tooling, libraries, or published models.

Responsibilities

• Own the Model Adaptation Lifecycle: Drive end-to-end adaptation of SLMs to specialized domain tasks, choosing the right base models, training strategies, and architectures.
• Apply Efficient Fine-Tuning: Implement parameter-efficient fine-tuning (PEFT) methods with a mathematically grounded understanding of low-rank adaptation (LoRA, QLoRA, rank selection, parameterization, scaling, and accuracy–efficiency trade-offs).
• Select Strategy per Task: Determine and sequence optimal adaptation strategies - including Continued Pre-Training (CPT), Supervised Fine-Tuning (SFT), and RL/preference tuning (DPO, GRPO, RLHF).
• Curate & Synthesize Data: Clean, curate, format, and generate high-quality datasets, including synthetic data generation when domain data is scarce.
• Design Functional Evaluations: Build task-specific benchmarks grounded in functional, executable correctness rather than loss curves or perplexity alone, maintaining regression tracking across versions.
• Leverage Knowledge Transfer: Use larger models as teachers to transfer capabilities to SLMs via distillation, structured pruning, and quantization.
• Deploy & Optimize Serving: Stand up and optimize inference infrastructure across GPU, CPU, and edge targets using frameworks such as vLLM, TGI, Ollama, llama.cpp, or TensorRT-LLM.
• Pipeline Reproducibility & Research Execution: Break target tasks into milestones with clear criteria, maintaining reproducible pipelines with data/config versioning, checkpointing, and experiment tracking.
• Translate recent literature into reproducible experiments.

Benefits

• Employment-based cooperation
• Comprehensive insurance for you and your family (health, life, and accident)
• Paid PTO policy (vacation, sick leaves, and public holidays)
• Tech equipment provided
• Udemy courses, workshops, trainings & expert knowledge-sharing
• Flexible hours & work setup