Narayan Jangid
Applied AI/ML Engineer · Bengaluru
I'm an Applied AI/ML Engineer. Professionally, most of my work has been on NLP and LLM systems: agent workflows, MCP servers and clients, retrieval over enterprise data, and fine-tuning and serving the models behind them.
My interests go well beyond what I've worked on so far. I spend a lot of my own time learning other parts of AI, and I write up what I study in my posts.
- At work
- NLP LLM systems Agents & MCP Retrieval & RAG
- Learning
- Time series & forecasting Inference engineering Reinforcement learning Embodied AI and more
Bengaluru, Karnataka
experience 3 yrs 5 mos in total
- Apr 2026 – Present 7 mos
Recruiterflow
Data Engineer (AI Infra)
- Architected the Natural Language Search data pipeline, processing 200K+ candidates and contacts and 10K+ jobs and deals a day to power LLM-driven search.
- Migrated the company-enrichment pipeline off a legacy system, adding caching and request optimizations that cut API costs by over 2.5×.
- Mar 2025 – Apr 2026 1 yr 1 mo
Kloudspot
Machine Learning Engineer
- Led the end-to-end design of LISA AI, an enterprise AI assistant, and managed and mentored a 5-member AI team.
- Benchmarked inference strategies and tuned vLLM/SGLang with EAGLE-3 speculative decoding for up to 2× throughput and 2.7–3.5× lower latency.
- Jun 2023 – Feb 2025 1 yr 9 mos
Celebal Technologies
Associate Data Scientist Nov 2023 – Feb 2025
- Led Azure OpenAI and Document Intelligence solutions to extract, embed and index enterprise documents, improving BLEU by 15%.
- Designed a hybrid edge–cloud VLM inference system for real-time video analytics, reaching 77.9% mAP.
Data Science Intern Jun 2023 – Oct 2023
- Fine-tuned GPT-3.5-turbo and Mistral 7B for text-to-query generation (SQL, NoSQL), with DataCompy scores above 95%.
- Improved a RAG pipeline over tabular data, raising search-result accuracy from 60% to 85%.
education
2017 – 2022