What AI Lever Shipped
From Oct 2024 to Jan 2026, AI Lever operated as a one-person studio — doing customer discovery, technical validation, and building AI products across enterprise and consumer use cases.
Chorus — The nervous system for your organization
Federated AI agents conduct structured conversations on behalf of decision-makers, capturing collective intelligence at scale while preserving anonymity. Built for organizations that need to learn and adapt faster than traditional facilitation allows.
Stack: LangGraph, LLM orchestration, structured output pipelines
SkyEye — Aerial Property Intelligence
Combines satellite imagery with vision-language models to detect and analyze property-level features at any address. Built for service businesses and enterprises needing scalable property intelligence without manual site visits.
Stack: VLMs, Azure Maps satellite imagery, object detection, segmentation
Plate360
iOS nutrition tracking using ARKit volumetric measurement and on-device ML to estimate food weight from a photo, plus LLM-based nutrition extraction. Deployed to live users.
Stack: ARKit, iOS, on-device ML, LLM extraction
AI Telephone Agent for Home Services
Inbound triage, Q&A, and booking agent for home service businesses. Handles structured knowledge extraction from unstructured business content, configurable escalation rules, and live call routing.
Stack: Voice AI, RAG, configurable knowledge extraction
The Founder
Sesh is an applied AI engineer and technology leader with 10+ years building intelligent software products — taking ML from prototype to production in search, GenAI, and healthcare automation, reaching hundreds of millions of users.
At Google Search and Brain, he was the technical lead for Google's first GenAI Search Summaries feature — recognized with a Spot Bonus for org-level impact. He built hybrid RAG pipelines, multi-stage safety filtering, and an LLM-as-judge evaluation framework adopted across 10+ Search verticals.
At AKASA, as founding ML engineer, he built the MLOps platform, human-in-the-loop architecture, and a universal field extraction system that reduced labeling costs by 70% and enabled new healthcare clients to go live in days instead of weeks.
At Walmart Labs, he co-led relevance and multi-tenancy architecture for ecommerce search across walmart.com, samsclub.com, and grocery.walmart.com — delivering learning-to-rank pipelines and revenue lift validated via A/B testing.
He holds 2 patents, has 5 international publications (including ICML 2020), and is a visiting scholar alumnus of CMU's School of Computer Science.
How I Think About AI
Just as a lever amplifies effort, the right AI architecture amplifies what a team can build. I've spent my career finding that leverage point — whether engineering Google's first GenAI Search Summaries for hundreds of millions of users, building foundational MLOps infrastructure from scratch at a healthcare startup, or shipping my own products as a founder.
I believe the best AI engineers work across the full stack: problem framing, data, modeling, evaluation, safety, and production. That's the lens I bring to every team I join.
Sesh is exploring his next chapter. If you're building something ambitious with AI and want a full-stack ML engineer with a founder's perspective — get in touch.