Forward Deployed AI Engineer · AI/ML Consultant · Systems Architect
I'm Jai Kumar Singh. I spent a career as a backend and full-stack architect at Expedia, Reliance Jio, Guavus, IBM Research and Oracle. Today I embed with teams to take AI systems from demo to dependable: RAG pipelines, agentic workflows, voice agents and the infrastructure that keeps them honest.
01 — About
Foundation · 2009 — 2025
Distributed systems, big data and real-time platforms at scale: Expedia's booking data ecosystem, Reliance Jio's cloud monitoring and autoscaling, petabyte analytics at Guavus, cloud management at IBM Research, clusterware at Oracle. As Chief Architect and CTO I owned technical direction end to end, and I have shipped machine learning to production since 2013.
Now · 2026 —
I work as a solopreneur, embedded inside client teams, on the last mile of AI: production-grade RAG pipelines, agentic workflows, voice agents, evaluation and LLMOps. Current clients are Pythagoras AI and RankUp. The systems background is the point. Models are the easy part; data contracts, latency budgets, failure modes and observability are where deployments succeed or quietly fail.
02 — Current work
All current engagements are delivered through Jai Kumar Singh’s Software Consultancy Services (GSTIN 06CIKPS5176H1ZR), a solopreneur practice run remotely from Gurgaon, India, with client visits where the work needs them.
Consulting on the engineering behind Clara, a fully managed AI care agent that holds clinical and administrative conversations by voice and text in English and Spanish. One agent, many use cases: appointment reminders, scheduling, health surveys and more, with structured data written back to the EHR and intelligent escalation to humans.
Advising on architecture and technical direction for RankUp's end-to-end solutions, which are agentic by design: answer-engine and generative-engine optimisation (AEO/GEO) that makes brands the primary cited answer in ChatGPT, Perplexity, Gemini and Google AI Overviews, plus commerce and customer acquisition.
A conversational assistant that runs infrastructure diagnostics such as ping, traceroute and DNS from spoken commands. Whisper and Sarvam for speech, real-time CLI execution and synthesised spoken responses.
Net-worth analytics that ingests CDSL and CAMS statements plus manual data and derives asset allocation and exposure across direct and mutual-fund holdings at source, fund and portfolio level.
Hands-on LoRA and QLoRA fine-tuning and MLX training on a 16 GB MacBook M1, to understand what small, local models can do before recommending a bigger bill.
03 — Earlier work
Led design and implementation of an AI-driven orchestration framework and intelligent Building Management System, automating complex workflows with ML pattern recognition. Built generative pipelines that produce building models and integrate with Revit and Tekla.
Owned the design of mapping and geospatial services on open-source systems. Designed graph-based locality resolution for Indian address normalisation and built routing infrastructure that uses traffic data and vehicle profiles to improve ETA accuracy.
Architected and evolved the booking data platform, aligning technical direction with business needs. Built scalable pipelines and domain adapters for enrichment and interoperability, enabling predictive profit estimation used in marketing bidding strategies.
Designed and implemented monitoring, alerting and autoscaling services as foundational cloud infrastructure, with visualisation and control interfaces for real-time observability and configuration.
Owned technology and platform development end to end. Built a real-time auction platform supporting high concurrency with low-latency bidding over WebSockets.
Multi-class SVM classification for web content categorisation and outlier detection on time-series network data, in production from 2013. Contributed MultithreadedTableMapper to Apache HBase (HBASE-5166).
04 — Experience
Founder · Forward Deployed AI Engineer & AI/ML Consultant, Systems Architect
*Client visits if required.
Live4est Inc · Remote
Pythagoras AI · Remote
Live4est Inc · Gurgaon
Intents Mobi Inc · Gurgaon
Expedia Inc · Gurgaon
Reliance Jio Cloud · Gurgaon
Blue Villa Trade and Auctions Pvt Ltd
Guavus Inc · Gurgaon
Oracle · Bangalore
05 — Technical expertise
06 — Background
Patent · Jan 2012
US 9,098,379 B2 · IBM
Publication · ICSEM 2011
Publication · IEEE SOCA 2010
IBM Research
07 — FAQ
I'm a forward deployed AI engineer, AI/ML consultant and systems architect. I work as a solopreneur, embedded with client teams, to take AI systems such as RAG pipelines, agentic workflows and voice agents from demo to production. My current clients are Pythagoras AI (AI/ML consulting) and RankUp (Technical Advisor).
A forward deployed engineer (FDE) works inside a customer's team and environment rather than from a vendor's office, integrating a product or AI system with the customer's real data, workflows and constraints until it delivers value in production. I combine that role with AI/ML consulting.
Sixteen years as a backend and full-stack architect: Principal Engineer at Expedia, Senior Manager at Reliance Jio Cloud, Chief Architect at Intents Mobi, CTO at BidsVilla, Lead Analytics at Guavus, and earlier roles at IBM India Research Lab and Oracle. B.Tech/M.Tech in Computer Science from IIT Kanpur. Machine learning in production since 2013, one patent, and a contribution to Apache HBase.
Forward-deployed AI engineering, AI/ML consulting, technical advisory roles and hard backend or data-platform architecture problems. Engagements usually start with a 30-minute call to frame the problem.
Gurgaon (Gurugram), India. I work remotely with clients worldwide. The best way to reach me is to book a 30-minute call, email me or message me on LinkedIn.
08 — Contact
Open to forward-deployed engagements, advisory roles and genuinely hard architecture problems. Most engagements start with a 30-minute call to frame the problem.