Himanshu DongrePune, India

Applied AI systems, from perception to agent infrastructure.

Associate Technical Architect working across automotive perception, embedded deployment, and reliable agent systems.

Portrait of Himanshu Dongre under flowering branches.

I architect production AI systems across autonomous driving, in-cabin intelligence, and constrained embedded platforms, connecting model feasibility to runtime design and target-hardware validation.

My independent work explores coordination, state, trust, and evidence-grounded decision support for agents through Smriti, Sentinel OS, and PropOps.

Personal projectsBackground
12+ yearsbuilding production software systems
7+ yearsbuilding and deploying AI systems
Portrait of Himanshu Dongre under flowering branches.

Current work

  • Multimodal perception and in-cabin intelligence under embedded constraints.
  • Agent coordination, state, evaluation, and human approval.
  • MLOps and DataOps architecture for repeatable experimentation and validation.
Experience

Experience.

The production systems behind the independent work.

2019 - PresentBengaluru / Tokyo / Pune

Associate Technical Architect · KPIT Technologies

Architects AI systems across automotive perception, in-cabin intelligence, model industrialization, and the platforms that carry them into production.

  • Leads technical direction from feasibility and architecture through C++ integration, optimization, and validation on target hardware.
  • Defined a multi-model in-cabin system spanning whole-body pose, gaze, and temporal action and skeleton forecasting on NVIDIA Jetson.
  • Developed reusable deployment patterns across NVIDIA DRIVE and Jetson, Qualcomm Snapdragon Ride, Ambarella, and Renesas platforms.
  • Enabled deep-learning workloads on constrained automotive ECUs through runtime selection, custom operators, quantization, graph optimization, and Python-to-C++ redesign.
  • Built perception middleware and vehicle interfaces across camera, LiDAR, radar, traffic-light recognition, tracking, and concurrent inference.
  • Leads MLOps and DataOps architecture, and advises internal teams on agent use cases, evaluation, orchestration, tool integration, state and memory, and human approval.

Earlier work

  • Software Engineer · Independent / Freelance Work2017 - 2018 · Bengaluru
  • Software Engineer · Accenture2014 - 2017 · Bengaluru
Full background
Projects

Personal projects.

Open-source systems built to test specific ideas in public.

Multi-agent coordination2026Active

Smriti

Version control for project reasoning state across coding agents.

multi-agent systemsreasoning statecoordination
Read notesWebsite
Evidence-grounded agent2026Public prototype

PropOps

An open-source research agent that joins government property records into an inspectable diligence workflow.

agent workflowsgovernment dataentity resolution
Read notesRepository
Trust kernel for agents2026Active

Sentinel OS

A local-first trust kernel for agent systems with policy checks, append-only history, and verification.

trustworthy agentspolicy enforcementcryptographic verification
Read notesRepository
Interests

Questions I keep returning to.

Questions that connect deployed perception, agent systems, and independent research.

Perception and multimodal intelligence

How models combine visual, temporal, and sensor evidence while meeting the constraints of deployed systems.

Reasoning state and coordination

How decisions, assumptions, and active work are represented, restored, and shared across agents.

Evaluation and trust

How agent systems expose evidence, policy, and execution history in forms people can inspect.

Decision support under uncertainty

How systems help judgment without pretending certainty where the evidence is incomplete or noisy.

I am most interested in intelligent systems when they become concrete: how state is represented, what can be verified, where interfaces fail, and how useful behavior survives real constraints.

Approach

Working defaults.

A few defaults that shape how I build and evaluate systems.

  • First principlesThe work should still make sense after you strip away tooling, hype, and surface detail.
  • Implementation as a testBuilding is part of thinking. Weak assumptions usually show up faster in systems than in discussion.
  • Clarity over theatreI care more about what survives scrutiny, constraints, and deployment than what merely performs well in a demo.
Ideas should survive code, data, and deployment.

I care about the loop between ideas and implementation. If an idea gets weaker once it touches code, constraints, interfaces, or data, it probably was not strong enough yet.

Log

Project log.

A short public record of releases and meaningful project changes.

May 2026Release

Smriti: coordination beyond handoff files.

Added active claims, stable task IDs, freshness checks, repository drift detection, worktrees, 21 MCP tools, and a local-first quickstart.

NotesWebsite
Apr 2026Prototype

PropOps: evidence-grounded property research.

Expanded the public prototype across 19 workflows, state RERA and IGRS sources, eCourts, builder entity resolution, and batch evaluation.

NotesRepository
Feb 2026Milestone

Sentinel OS: MCP enforcement proxy.

The v0.3 prototype intercepts filesystem MCP calls, applies default-deny policy, and records signed transitions for offline verification.

NotesRepository
Links

Contact.

The best entry points are the projects, repositories, and direct email.

Reach out or follow along

Email is the easiest way to reach me. You can also find current public work through GitHub, LinkedIn, X, and Instagram.

EmailDownload CV (PDF)

Public links

GitHub is the best place to start for current public work. LinkedIn and X carry occasional public updates and context.