About — Yadvendra Kshatri

I've spent the last several years building and scaling AI products, from NLP research prototypes to production agentic systems running for enterprise clients. I co-founded Aidetic in 2020 and grew it from zero to a 65-person, cross-functional organization, doubling revenue year-over-year for three consecutive years before our acquisition.

Today I lead pre-sales and forward-deployed engineering across all of our AI/ML engagements; embedding with client engineering teams to scope, prototype, and ship LLM-powered systems. I care deeply about agent architectures, evaluation frameworks, context engineering, and the cost and latency tradeoffs that decide whether an LLM project actually makes it into production.

Experience

Head of AI & Co-founder

Formerly Aidetic Software Now, Aays Analytics

Mar 2020 — Present · Bengaluru

Co-founded and scaled Aidetic from 0 to a 65-person cross-functional organization; established hiring, engineering, and AI delivery processes from scratch.
Drove 100% year-over-year revenue growth for three consecutive years and ensured a clean handover post-acquisition.
Lead pre-sales and forward-deployed engineering on every AI/ML engagement, advising on agent architectures, evaluation, context engineering, and cost optimization.
Lead the strategic Databricks partnership: authored training curricula for 70+ team members, ran 40+ client workshops and technical deep-dives, and shipped reusable docs and sample code (including a migration accelerator that speeds competitive migrations by ~40%).
Architect and ship production agentic and LLM-powered systems for enterprise clients; RAG/retrieval pipelines, tool-use orchestration, and evaluation harnesses with ~75% user acceptance, <2.5 min latency for ~11 LLM calls and tool executions, saving ~1,800 hours/day.

Head of AI & Co-founder · Fabric Analytics

Jul 2023 — Aug 2024 · Bengaluru

Architected the AI and data science engine for a D2C e-commerce platform.
Designed a unified database layer integrating Shopify, Google Ads, and Amazon, with RAG and prompt-chaining infrastructure powering text-to-SQL.
Built AI-driven anomaly detection, root-cause analysis, and a chat assistant tuned for the D2C industry.

NLP Engineer · Nesh

Feb 2021 — Feb 2022 · Bengaluru

Built enterprise-grade Q&A systems with domain-specific knowledge integration, document processing pipelines, and knowledge graphs from glossaries and customer documentation.

Data Analyst · Coviam

Dec 2018 — Mar 2020 · Bengaluru

Delivered data science initiatives for a leading Indonesian e-commerce platform — real-time license-plate detection, image-based recommendations, and CRM/promo analytics.

Data Analytics Associate · ZS Associates

May 2017 — Dec 2018 · Pune

Analyzed market research datasets to inform Fortune 500 client decisions; modelled consumer behaviour across 6 countries; upgraded VBA tools for conjoint analysis.

Selected projects

Agent Factory — Docomo Group retail company

Designed and shipped an "Agent Factory" platform letting 400 business users across 5 domains deploy 10+ LLM-powered agents. Owned agent architecture, tool-use design, prompting, and human-in-the-loop guardrails.
Ran a data-maturity assessment across 1,200 tables and built retrieval + reasoning pipelines that answer complex business questions over heterogeneous sources; ~75% user acceptance, <25k tokens per query.
Built the platform's evaluation harness; task-level success criteria, golden datasets, continuous-eval dashboards, that drove user acceptance from ~30% to ~75%.

Fintech — Monitoring and Valuation for Investment Funds

Built models to identify competitors from company descriptions and financial data, plus language models recommending companies for investment and competitive analysis.
Designed pipelines to extract and transform data from financial documents for downstream analysis.

Sales AI Assistant — Document Search

Built an AI research assistant ranking and summarizing internal documents for oil & gas, banking, and chemical clients; constructed knowledge graphs from unstructured data and open repositories.

OKR Quality Checker

Defined evaluation criteria with clients in a data-scarce setting and built a labeled benchmark of 500+ examples.
Shipped an MLOps pipeline with continuous evaluation, feature and performance drift monitoring, and monthly automated retraining with manual override.

Skills

Deep Learning & NLP: LLMs, RAG, prompt and context engineering, text-to-SQL, conversational AI, evaluation harnesses.
Infrastructure & tools: Python, PyTorch, Azure Databricks, OpenAI, Elasticsearch, Airflow, AWS, Azure, MLOps, Git.
Leadership: Project scoping, team management, client relationships, technical strategy, architecture design.
Analytics: Statistical modelling, CRM analytics, market research, business intelligence.

Education

Indian Institute of Technology (ISM), Dhanbad — B.Tech, 2013–2017.