Machine Learning Engineer

My name is Ankur Mishra, a B.Tech CSE student (RGIPT, 2026) with strong experience in LLM fine-tuning, deep learning, and research-driven model development. I am writing to explore ML/LLM engineering opportunities, especially in teams working on LLM development and AI platforms.

Key experience:

Fine-tuned Llama-based, RoBERTa, and code-generation models using QLoRA, PEFT & full fine-tuning

Built a hybrid Knowledge Distillation + LoRA system (LLM-NEO) for efficient Llama-3 model compression

Trained 10+ multilingual code-generation models using DeepMind datasets

Applied BF16 mixed precision, gradient accumulation, cosine LR scheduling, and curated datasets for stable & efficient training

Completed research on a multimodal, multi-task, multi-class classification system on a complex English dataset

Interned at Asynq Technologies and Coding Jr, working on LLM-based features, model optimization, and training workflows

Current focus areas:

Strengthening skills in MLOps (reproducible pipelines, deployment, evaluation frameworks, inference optimization)

Deepening expertise in quantitative ML, time-series modeling, and ML-driven decision systems

Building a long-term career in core ML/LLM engineering