Hi, I'm Abhigith Neil Abraham.
I build ML and software systems end-to-end: LLM inference optimization on the latest GPUs, coding-agent harnesses, large-scale data transformation and enterprise automation. I'm currently CEO of Vitalops.ai.
Over the past 6+ years I've worked across the entire ML, full-stack and large-data lifecycle: benchmarked LLM inference speed in production, built SOTA coding agents, shipped products to paying enterprise customers, published peer-reviewed research, and maintained open source used by thousands. Along the way I've mentored 5000+ people on their tech careers.
Outside of work I teach, write, travel and organise techno events — that side of me lives on the personal page. You can also grab the PDF resume.
Open source
- Agentic coding harness.
- Large-scale LLM-powered data transformation engine.
- Multi-machine computer-use MCP server.
- Natural language to SQL over tabular data.
- Distributed hyperparameter optimization across machines.
- Carbon and energy tracking for ML experiments.
Contributions
- Added
KeyPoints.merge(); fixedprocess_videohanging past the end of a video, out-of-bucket detections in size-bucketed metrics, crop/overlay annotators drawing outside the scene, and empty/numpy-index handling in key points. - Added the
TextEmbeddertask (sentence-transformer embeddings) and its docs. - Implemented
DistributeScan/RemoteScanfor distributed hyperparameter search across machines, later spun out as Jako.
Publications
- 15th International Conference on Data Science, Technology and Applications. A MAP / FILTER / REDUCE framework for LLM transformations on arbitrarily large datasets, with an agentic planner; 96.2% accuracy on large-scale enrichment.
- Biomedical sentence embeddings via contrastive learning, at a peer-reviewed clinical NLP venue.
- Paper accompanying the TableQA NL-to-SQL engine.
- Chapter in Advanced Deep Learning for Engineers and ScientistsPeer-reviewed engineering book covering deep learning across computer vision, biometrics and smart grids.
Writing
Talks
- An hour-long conversation on what ChatGPT changes and where it falls short.
- Presenting TableQA, natural-language querying over tabular data, at the FOSS United June meetup.
- Open Q&A on open source, ML and getting started in tech.
Skills
- LLM Inference & Performance
- Inference optimization, speculative decoding, quantization, GPU profiling & bottleneck analysis, confidential computing, load testing · vLLM, SGLang, NVIDIA Blackwell GPUs
- Agent Systems
- Agent harness design, multi-agent orchestration, verify-then-escalate chains, LLM-as-judge verification, tool & MCP design, context engineering, failure-mode analysis
- Evaluation & Research
- Eval harness design, benchmark replication, experiment design & ablation, statistical significance testing, accuracy-regression verification · Terminal-Bench, BFCL
- Systems & Delivery
- Distributed GPU infrastructure, full-stack engineering, large-scale data transformation, secure deployment & attestation · Python, SQL, Postgres, Docker, AWS, GCP, Linux
Education
- College of Engineering, Trivandrum — BTech, Electrical and Electronics8.1 / 10 CGPA.
Community
- Founded Tinkerhub CET; core contributor to The Fast Tracked Programmer5000+ people mentored on their tech careers.
- Free ML workshops across Kerala with Google AI Explore ML
- Teaching maths and science at GEMS School in the Himalayas
Contact
I'm best reached via email, LinkedIn or Twitter. Always open to interesting conversations and collaboration.