Dinesh Jinjala

Dinesh JinjalaAI/ML Engineer

I build AI systems end to end: machine learning models, LLM agents, and the Python, Rust and React software that runs them in production.

See the workAsk Dinesh

Scroll to take the stack apart

  1. 012015–17

    Copper

    Embedded Systems Engineer

    I designed PCBs for motor control: a BLDC motor driver, and a 5-axis stepper driver for a robotic hand.

    • BLDC motor driver
    • 5-axis robotic hand
    • PCB layout

    DJ-MC5 A modern redraw of the motor-control boards I designed in 2015–17: BLDC drive plus a 5-axis stepper driver for a robotic hand.

  2. 022015–17

    Firmware

    Embedded Systems Engineer

    I wrote the Embedded C that ran on those boards: timing-critical motor control on microcontrollers.

    • Embedded C
    • Microcontrollers
    • Motor control

  3. 032017–22

    Data and models

    Data Analyst & AI Scientist · Freelance ML Engineer

    I built ML models and data pipelines for clients worldwide: computer vision, NLP, deep learning and Spark.

    • Computer vision
    • NLP
    • Deep learning
    • Spark

  4. 042022–26

    Production platform

    Senior Software Developer → Staff → Senior Staff Software Engineer

    I was a lead engineer on a multi-tenant Pharma Manufacturing Analytics SaaS for 4+ years. It is used by 10+ enterprise pharma companies, each with several manufacturing sites.

    • SPC & capability analytics
    • Batch evaluation engine
    • Report platform
    • SSO & identity

  5. 052026–now

    Governed agents

    Senior Staff Software Engineer (Consultant)

    I build AI agents that are safe to run on real enterprise data, and fast systems in Rust. 5th of 362 in the ICML 2026 Agent Reproduction Challenge.

    • Governed ReAct agent
    • Need-to-Know
    • Vektor
    • RoadAI

From copper to cognition

ICML 2026 Agent Reproduction Challenge
5th of 362
enterprise pharma clients on the platform I helped lead for 4+ years
10+
freelance projects, 4.9★ average
160+
See the workAsk Dinesh
02

Selected work

Three systems up close, then the rest. Each one has a full case study: the problem, how I built it, and what changed.

01A multi-tenant Pharma Manufacturing Analytics SaaS

Governed AI Analytics Agent

Lead engineer

An AI analyst that answers quality questions in plain English over regulated pharma data. I replaced the first-generation agent with a governed design: the model works through validated tools, none of its SQL reaches the database unchecked, and a review step checks every answer before anyone sees it.

7/7
correct in a live side-by-side test (previous version 6/7)
~30%
faster mean response (~12 s vs ~17 s)
0
regressions

Python / FastAPI / PostgreSQL / LLM tool calling

Read the case study

02Client project

RoadAI: road design CAD

Sole developer

I built a web CAD platform for road design. Engineers import LiDAR point clouds as 3D terrain, design horizontal and vertical alignments with cut/fill analysis, find routes with A*, and get compliance reports against Indian road standards (IRC).

LiDAR → terrain
point clouds to 3D surfaces
A*
route finding over terrain
IRC
compliance reports against Indian road standards

Rust / Axum / React / TypeScript / Three.js

Read the case study

03A multi-tenant Pharma Manufacturing Analytics SaaS

AI Agent Platform & MCP Gateway

Architect and developer

I designed and built a platform where a new AI agent is a configuration file, not new code. Every tool call passes through one governed MCP gateway, and generated code runs in an isolated sandbox. Consolidating tools made agents faster and cheaper to run.

~15
tool integrations behind one gateway
30 → 1
tool calls in one workflow, after I replaced them with one bulk tool
Sandboxed
generated code runs in an isolated container

Python / MCP / Docker / YAML

Read the case study

All projects

  1. 01Governed AI Analytics AgentLead engineer · A multi-tenant Pharma Manufacturing Analytics SaaS7/7correct in a live side-by-side test (previous version 6/7)
  2. 02RoadAI: road design CADSole developer · Client projectLiDAR → terrainpoint clouds to 3D surfaces
  3. 03AI Agent Platform & MCP GatewayArchitect and developer · A multi-tenant Pharma Manufacturing Analytics SaaS~15tool integrations behind one gateway
  4. 04Multi-Tenant SSO & IdentityLead developer · A multi-tenant Pharma Manufacturing Analytics SaaS2.65 s → 0.44 spassword sign-in, ~6× faster
  5. 05SPC & Process Capability AnalyticsBuilt and owned the engine · A multi-tenant Pharma Manufacturing Analytics SaaS13chart types (I-MR, EWMA, Pareto…)
  6. 06Automated Report PlatformLead developer · A multi-tenant Pharma Manufacturing Analytics SaaS1 data layerevery report draws from it
  7. 07Batch Evaluation & Export EngineBuilt v1, main developer of v2 · A multi-tenant Pharma Manufacturing Analytics SaaS~4 GB → ~80 MBmemory to export 1M+ rows (development benchmark)
  8. 08Vektor: code context for AI agentsAuthor, open source · Personal project1.000Recall@5 on my 20-query benchmark
  9. 09SQLite Graph AcceleratorAuthor, open source · Personal project26×faster than indexed SQLite at depth 6
  10. 10Need-to-Know: a data agent built not to leakAuthor, open source · Personal project5MCP tools, none returns raw rows
  11. 11Manufacturing Failure PredictionAuthor, public repo · Personal research590sensor signals per batch
  12. 12Quant Research PlatformAuthor, private · Personal research157K+instruments searchable
  13. 13No-Code AutoML Pipeline BuilderCore contributor · A multi-tenant Pharma Manufacturing Analytics SaaSNo codeengineers draw the pipeline as a graph
03

5th of 362

274 reproduction logbooks · ICML 2026 Agent Reproduction Challenge

My agent, DineshAI, checked the claims of ICML 2026 papers against their text and code, and logged the evidence for every one.

Browse every paper
04

What clients said

From my freelance years: 160+ projects, 4.9★ across 130+ ratings, clients in 20+ countries

  • “Dinesh is a very good Python programmer who has a good knowledge of data science. He is very patient and communicates well in English. I strongly recommend him as he can be an technical asset to complete any sophisticated data science project.”

    Hassan D. Yong-dong, KR

    Help needed for an ongoing machine learning project · May 2022

  • “Dinesh worked for me on couple of project related to machine learning , NLP and deep learning. he was very diligent and meticulous in his work. Amazed by his work and knowledge in the field of data science, brilliant programming skills in python. …”

    Bhanu S. Data Scientist

    Model Creation for machine learning · Mar 2020

  • “Dinesh has a great skills in Python, Spark and Bigdata. I really like the speed of Dinesh. He has finished my project before my deadline. I really recommend him for Python, ML, DL, Bigdata, spark and Hadoop.”

    Adel C. Egham, GB

    I need to solve big data projects · May 2020

  • “… there are few people that can really get into your project and understand your requirements. He is one of these few, he is hardworking and I'm happy that I found him! … TRUST ME AND GIVE THIS MAN SOME WORK!! You'll be surprised :)”

    Christian C. Bucharest, RO

    Python project · Jun 2021

  • “Dinesh is professional, knowledge of Python is exceptional, code is written elegantly, easy to understand.”

    Tyrone C. Cape Town, ZA

    Python project to be completed · Apr 2020

  • “He continues to deliver good results whenever I need his expertise. He is very reliable for long term assignments. Keep up good work Dinesh.”

    Ramya R. Los Angeles, US

    Need coding help and mentoring · May 2019

  • “Great developer who really cares about his clients and always delivers. Highly recommended.”

    Colin N. Bournemouth, GB

    Assistance · Sep 2018

  • “It was a great experience working with Dinesh. He was easy to work & communicate with. He brought some good ideas to the project. Always give prompt reply and always give accurate solution. Would highly recommend him to anyone.”

    Upwork client 75-hour contract

    Python and Shell programming

05

Ask Dinesh

An AI assistant that knows my projects. Ask about my work, or leave a short brief and it comes straight to my inbox.

Assistant

Ready

  1. assistant

    Hi. Ask me anything about Dinesh's projects, skills or availability.

An AI model writes these answers. Contact details you share are emailed to Dinesh.