Founding AI Engineer at Observo AI, now part of SentinelOne. I led the 0→1 build of Observo’s AI stack, mentoring a team of 5 across AI, backend, frontend and DevOps, and worked with sales and marketing to drive 1M+ ARR.
I built low-latency PII and log-schema detection, fine-tuned LLMs for log intelligence (CIM/OCSF ingestion, anomaly detection, summarization, Grok detection), and Orion, an agentic chat-copilot for security researchers that enterprise customers adopted.
Senior Software Engineer and Tech Lead on the Content Abuse AI team at LinkedIn. We built multi-modal defenses that prevent 5M+ spam views a week: fine-tuned multi-modal models, agentic frameworks and in-house pre-training for domain adaptation, over datasets of 10M+.
Highlights: a multilingual end-to-end OCR stack served through Samza and Kafka, video classifiers for spam, and Flash, an in-house framework for scalable LLM training and deployment. I mentored 5+ engineers.
Senior Research Engineer in the Augmented Reality team at Samsung Research Institute Bangalore. I built on-device AR/VR for premium camera apps, covering semantic segmentation, multi-modal detection and tracking, scene understanding and multi-view 3D reconstruction, all running on Android.
Projects include Quick Measure, AR Doodle, camera-based human height measurement and single-shot 3D reconstruction, which helped shape the early AR stack on the Galaxy S and Note series.
Also
Pro bono: I designed the fintech intelligence layer for a personal-finance startup, built on the account aggregator framework: automatic spend categorisation, a conversational finance copilot, credit card statement parsing and keyword-based transaction search.
Skills: PyTorch and TensorFlow; agentic frameworks (LangChain, GraphRAG, CrewAI, RAG); MLOps (MLflow, DVC); large-scale datasets from research to production; leading AI projects and mentoring engineers.