Resume

Production ML engineering across generative AI, image-processing systems, model adaptation, and scalable inference.

• Fine-tuned and quantized diffusion models using LoRA for image editing and unblurring pipelines.
• Integrated VLM/LLM systems for automated image description, prompt refinement, and image-processing workflows.
• Deployed scalable FastAPI-based inference services for production ML applications.
• Built model optimization pipelines involving LoRA, quantization, and production-oriented serving constraints.

PyTorch · Diffusers · Transformers · LoRA · FastAPI · vLLM · Docker

Kaggle Notebook Master & Competition Expert

Master rank in Kaggle Notebooks and Expert rank in Kaggle Competitions.

3rd Place & Best Dataset Award — DL Sprint 3.0 | Bengali AI Math Olympiad

Developed an AI pipeline for Bengali math problem solving with 84% overall accuracy.

20th Place — Make Data Count: Finding Data References

Developed a multi-stage pipeline for detecting dataset mentions and classifying dataset references in research papers.

18th Public / 37th Private — AI Mathematical Olympiad Progress Prize 2

Built an LLM-based pipeline for national-level math problem solving.

25th Public / 26th Private · Silver Medal — Drawing with LLMs

Built an end-to-end constrained image generation and image-to-SVG pipeline.

ICPC Asia Dhaka Regional Preliminary Contest 2024

10th / 2,489 teams

ICPC Asia West Continent Final 2023

Finalist

ICPC Dhaka Regional Rankings

Ranked 14th in 2023, 24th in 2021, and 29th in 2020.

Codeforces Expert

Highest rating 1798; solved 3000+ problems across platforms.

Contest Contributions

Contributed to multiple programming contests as a problem setter, developer, tester, and judge.