Hello! I’m Abhijit More, a Machine Learning Engineer at F-Secure (formerly Sony AI/ML).

I build production systems for agentic AI, RAG, low-latency inference, and search and ranking. Most of my work spans the full stack: retrieval pipelines, model serving, evaluation, and the infrastructure that keeps systems reliable in production.

What I Work On

At F-Secure, I work on production ML systems for user safety and threat intelligence. Before that, at Sony, I focused on edge inference and computer vision, including distributed transformer deployment on Jetson and Raspberry Pi and image QA pipelines on Jetson Orin NX.

Earlier, I trained at IITM Pravartak on applied LLM and transformer work, and started my career at HighRadius building ML-driven financial intelligence systems.

Open Source

My public projects focus on problems I care about in production ML:

• RelevanceIQ — hybrid retrieval, reranking, and citation-grounded RAG over 568K reviews
• KAIROS — decision intelligence with hybrid ensembles, calibration, and observability

Both are measured end to end, not demo-ware. Write-ups are on this site under Projects.

How I Think About the Work

I start with a clear problem and a strong baseline. I care about serving latency, retrieval quality, and observability from day one, not after a model looks good offline.

Always open to connecting about applied ML, search, RAG, agents, or production inference.