anmol gautam

I am a Lead Applied Scientist with 4+ years of experience building and shipping production-grade AI systems across multi-agent platforms, enterprise RAG, Text-to-SQL, LLM fine-tuning, multimodal AI, and autonomous coding systems. My work combines applied research, product engineering, and scalable AI infrastructure, with a strong focus on taking complex AI ideas from prototype to real-world deployment.

Currently, I work as Lead Applied Scientist - AI/ML at 8bit.ai, where I architected Neutrino, a multi-agent AI platform for enterprise search, Text-to-SQL, and workflow automation. The platform uses FastAPI/SSE, human-in-the-loop execution, observability, and multi-LLM orchestration, and has been deployed across multiple major ISV partners. I have also worked on domain-specific LLM fine-tuning using LoRA, DoRA, PEFT, DPO, and GRPO, along with inference optimization using vLLM, SGLang, pruning, quantization, and benchmarking.

Previously, at SuperAGI, I built RAG-based and Text-to-SQL conversational multi-agent systems and contributed to SuperCoder 2.0, an autonomous code navigation and issue-resolution system that achieved 33% on SWE-Bench-Lite. I also worked on SAM-7B, an instruction-tuned Mistral-7B model built with custom datasets and evaluation pipelines, achieving GPT-3.5-comparable performance on selected benchmarks despite being trained on a smaller dataset.

Earlier in my career, I worked at Oracle, NVIDIA, and Gahan AI on document AI, information extraction, NLP, computer vision, machine translation, object detection, image segmentation, and video classification. My technical background includes PyTorch, Hugging Face, Transformers, FastAPI, PostgreSQL, Trino, MongoDB, ChromaDB, Milvus, pgvector, Docker, AWS, Azure, MLflow, vLLM, and SGLang.

Academically, I completed my M.Tech in Computer Science and Engineering from NIT Meghalaya with a 10.0 CGPA, receiving the Gold Medal in Academics and the Institute Best Master’s Thesis Award. I have authored/co-authored six publications across IEEE, Springer, and arXiv.

My core interests are reliable agentic AI systems, LLM post-training, enterprise AI workflows, AI copilots, retrieval systems, reasoning/evaluation pipelines, and scalable AI infrastructure. I enjoy roles where I can combine deep technical ownership with product thinking: designing systems from first principles, experimenting with models and architecture, optimizing for real user outcomes, and shipping AI products that work reliably in production.