# Pankaj Rajan

> Pankaj Rajan is Head of AI & Engineering at Eightfold AI and a two-time founder who builds agentic AI, search, and language systems used by millions. He makes time for occasional sessions with founders, leaders, and investors.

Pankaj Rajan is Head of AI & Engineering at Eightfold AI, based in the San Francisco Bay Area.

I have spent 15+ years building AI that real people use — and the last decade leading the teams and companies that build it.

At Eightfold AI, I lead AI and engineering for the company’s agentic products: the shared platform agents run on, and the agents, assistants, and analytics built on top of it. I have also built multi-agent orchestration for shopping and customer support for one of the premier frontier AI companies. Earlier, as co-founder and CPTO of MarkovML, I raised $8M and took an enterprise AI platform to SOC 2 Type II. Before that, I led conversational AI at Amazon Alexa, founded the Query Understanding team behind Walmart.com search, and was the second engineer on eBay’s machine translation engine.

I write about what I have learned building these systems, and I make time each month for a few sessions with founders, leaders, and investors working through the same problems.

## Systems built
- **Multi-agent shopping & customer support** (Multi-agent orchestration): An orchestrator that routes each conversation across specialized agents — product discovery, recommendations, orders, and support resolution — built for one of the premier frontier AI companies.
- **Enterprise agent platform** (Agents · Platform): A shared runtime and authoring layer for agents — orchestration, memory, tool access, guardrails, tenant isolation, and a built-in evaluation layer — so several product lines ship agents on one foundation instead of one stack per team.
- **Conversational workforce analytics** (Workforce intelligence): An agentic analyst for HR and business leaders: plain-language questions about skills, readiness, and talent become answers with the analysis behind them — no analyst queue.
- **Semantic search at e-commerce scale** (Search · NLP): Query understanding — entity extraction, intent classification, and query rewriting — for one of the largest product catalogs in the world, with +85% long-tail understanding.
- **Open-domain conversational AI** (Conversational AI): Generation, retrieval, relevance, and safety models for a consumer voice assistant, doubling user satisfaction while cutting cost to serve ~15%.
- **Machine translation for global commerce** (NLP · Translation): A production translation engine that opened cross-border marketplaces to buyers in their own language, adding $1B+ in GMV.

## Selected results
- **One agent platform for every product line** — Eightfold AI. Architected a shared agent platform — runtime, authoring, memory, tool access, guardrails, and evaluation — and took agentic products from first code to production on it. (One platform, many agents)
- **Semantic search for Walmart.com** — Walmart Global Tech. Founded the Query Understanding team — entity extraction, intent classification, query rewriting — and led the Endeca replacement across a 100+ engineer organization. (+85% long-tail · −13% infra cost)
- **An enterprise AI platform, built from zero** — MarkovML. Delivered model routing, retrieval, fine-tuning, inference, and agent workflows on multi-tenant infrastructure certified to SOC 2 Type II. (Deployment: 1 week → hours)
- **Higher quality at lower cost to serve** — Amazon Alexa AI. Redesigned generation, retrieval, relevance, and safety systems, with a new evaluation process gating every model change. (2× satisfaction, −15% cost)

## Expertise
- **Applied Domains:** Workforce intelligence & talent; Conversational analytics; Hiring & career agents; Shopping & customer support agents; E-commerce search & discovery; Voice & conversational assistants; Defense & EdTech AI
- **Agent Systems:** Agent runtimes & harnesses; Orchestration & authoring tools; State, context & long-term memory; MCP & tool interfaces; Multi-agent coordination; Long-horizon workflows; Human-in-the-loop checkpoints; Guardrails, permissions & tenant isolation
- **NLP & Conversational AI:** Entity recognition & linking; Intent classification; Machine translation; Open-domain dialogue & response generation; Response relevance models; Knowledge graphs; Multimodal assistants (voice + text)
- **Search & Ranking:** Query understanding & rewriting; Semantic & hybrid retrieval; Learning-to-rank; RAG over proprietary data; Long-tail query handling; Offline retrieval & ranking evaluation; Search platform migrations at scale
- **Classification & Evaluation:** Text & intent classifiers; Safety & policy classifiers; Agent & retrieval evals; Quality scoring & human review; Release gating & regression tracking; Production observability
- **Models & Platform:** Model routing: frontier, open-source, private; PEFT / LoRA fine-tuning; vLLM inference; Python · PyTorch · Kubernetes; AWS · GCP · Azure; SOC 2 Type II · FedRAMP-aligned

## Experience
- **Head of AI & Engineering, Eightfold AI** (2025 – Present): Lead AI and engineering across talent management, agentic candidate experiences, career coaching, talent design, and workforce readiness — products used by Fortune 1000 enterprises. Lead the evolution of the agentic platform, combining reasoning models, retrieval, grounding, tool use, predictive ML, and evaluation, and set technical direction across orchestration, enterprise integrations, security, tenant isolation, data residency, and production reliability.
- **Co-Founder & CPTO, MarkovML** (2021 – 2025): Led a 35-person company with engineering, platform, security, and go-to-market reporting in. Raised $8M, sold into defense and EdTech, achieved SOC 2 Type II, and presented strategy and risk to the board. Merged in 2025.
- **ML Lead, Conversational AI, Amazon Alexa AI** (2020 – 2021): Built early LLM and retrieval-augmented architectures for open-domain conversation, with work on safety alignment, response ranking, and knowledge graphs. Doubled user satisfaction, reduced cost to serve by ~15%, and delivered the evaluation system that gated every model change before production.
- **Co-Founder & CEO, Ingenus** (2018 – 2020): Raised pre-seed funding and launched a multimodal conversational shopping assistant. Wound down when COVID closed retail, returning remaining capital to investors.
- **ML Architect, Search Science, Walmart Global Tech** (2015 – 2018): Founded the Query Understanding team behind Walmart.com semantic search, improving long-tail understanding by 85%. Led the replacement of Oracle Endeca across a 100+ engineer organization, reducing infrastructure cost ~13%.
- **ML/NLP Engineer → ML Architect & Lead, Ayasdi AI · eBay · Microsoft · AMD** (2009 – 2015): At Ayasdi AI (2014–2015), architected and built the supervised machine learning platform on topological data analysis — unlocking several million dollars in new revenue in its first year across healthcare, finance, and fraud detection. Earlier, a core engineer on eBay’s machine translation engine ($1B+ added GMV), then Microsoft and AMD.

## Links
- Website: https://www.pankajrajan.com/
- LinkedIn: https://www.linkedin.com/in/pankaj-rajan
- Google Scholar: https://scholar.google.com/citations?user=-y7FX5wAAAAJ
- GitHub: https://github.com/pnrajan
- Digital twin (chat): https://www.twincafe.com/about?referral=KNdVJ-dy

- Email: pankaj@pankajrajan.com