Your Name
AI Research Engineer / Applied Scientist
City, India · +91 98XXX XXXXX · your.name@email.com · linkedin.com/in/yourname · github.com/yourname
Professional Summary
Research minded ML engineer with an M.Tech in Computer Science and 3 years training Deep Learning models in PyTorch. Fine tuned 7B Transformers with LoRA and PEFT for Indic language tasks, improving task accuracy by 18 points over prompting baselines. Reproduced two published papers end to end, with code and write ups on Hugging Face and GitHub.
Skills
- Deep Learning: PyTorch, Transformers, Fine tuning, LoRA, PEFT, QLoRA, Hugging Face (Transformers, Datasets, TRL), NLP
- Training at Scale: Distributed Training (DDP, FSDP), Mixed Precision, CUDA basics, Experiment Tracking (Weights & Biases)
- Maths: Linear Algebra, Probability, Optimisation, Statistics
- Engineering: Python, NumPy, Evaluation design, Git, Docker
Professional Experience
Machine Learning Engineer (Applied Research), Fabrikam AI Labs, Bengaluru Jul 2022 to Present
- Fine tuned 7B open Transformers with LoRA and PEFT for Hindi and Tamil summarisation; improved ROUGE scores by 18 points over prompting baselines.
- Built Evaluation pipelines combining automatic metrics and human ratings, with ablations that guided a 30% cut in training cost.
- Scaled training to 8 GPUs with Distributed Training (FSDP) and mixed precision, reducing epoch time from 9 h to 1.4 h.
- Tracked 400+ runs with Experiment Tracking in Weights & Biases, making results reproducible across the team.
Data Scientist, Contoso Retail Pvt. Ltd., Pune Jun 2021 to Jun 2022
- Built PyTorch demand forecasting models across 400 stores, reducing forecast error by 12%.
Projects
Paper Reproduction: LoRA vs Full Fine tuning on Indic NLP · PyTorch huggingface.co/yourname
- Reproduced the paper's main result within 1.2 points; ran 3 ablations (rank, target modules, data size) and published code, adapters and a write up.
Mini GPT from Scratch · PyTorch, CUDA github.com/yourname/MiniGPT
- Implemented multi head attention and a 25M parameter Transformer trained on 50MB of text; analysed scaling with model size.
Education
M.Tech, Computer Science (Machine Learning), Your University · 2021
B.Tech, Mathematics and Computing, Your University · 2019
Certifications
- Deep Learning Specialization, DeepLearning.AI
- Hugging Face NLP Course