booting santosh@portfolio...
[ok] loaded profile
[ok] mounted automotive-architecture modules
[ok] initialized ai-automation lab
[ok] ready
➜ ~ whoami
➜ ~ cat role.txt
Automotive Software Architect
12+ years architecting safety-critical vehicle software — IVI, telematics and in-vehicle networks for global OEM programs.
➜ ~ cat ai-lab.txt
AI Automation Expert AI LAB
In my AI lab: LLM agents, RAG pipelines and MCP tools — built hands-on, shared openly.
➜ ~ ls commands/
➜ ~
Automotive software architect with 12+ years designing and delivering high-performance embedded and vehicle software — IVI, telematics, and vehicle-network platforms — for global OEM programs including Lucid EV, Tata Motors, CNHi, Panasonic, and Rockwell Automation. AWS Certified Solutions Architect. I work from firmware and protocol stacks up to service architecture and validation toolchains, and lead cross-functional teams of 8–12 engineers from architecture through delivery.
Outside the architecture work, AI automation is my AI lab: LLM agents, RAG pipelines and MCP tools, many of them aimed at making automotive validation faster.
➜ ~ role --primary
➜ ~ role --ai-lab
"automotive_architect": {
Architecture and delivery of IVI, telematics and vehicle-network platforms — from protocol stacks and firmware to service design and validation toolchains.
Architecture & design
Vehicle networks & protocols
Platforms
Languages & fundamentals
Validation & delivery
Tooling
},
"ai_automation_lab": {
My AI lab: LLM agents, RAG pipelines, MCP tools and workflow automation — built hands-on and shared openly.
LLMs & agents
Retrieval & vector search
Automation
Build & serve
Machine learning
}
14 projects — shipped work, builds in progress, and what's coming next.
## problem
Engineers lost significant time manually searching ECU documentation and prior test cases to resolve engineering queries, slowing validation cycles.
## approach
Architected an LLM-powered chatbot combining RAG with a FAISS vector store over ECU/engineering docs, exposed through REST + gRPC for internal toolchain and CI/CD integration.
## result
0% reduction in query resolution time; improved documentation retrieval accuracy; deployed as a production internal tool.
## problem
IVI test suites required constant manual maintenance as interfaces evolved; writing new scripts from specs was slow.
## approach
Built an AI orchestration layer on Robot Framework that auto-generates and self-heals test scripts from natural-language specs, communicating with the execution engine over gRPC for real-time control and result streaming; integrated into TeamCity + GitLab CI/CD.
## result
Targeting ~0% reduction in test maintenance overhead through intelligent script repair and adaptive generation.
TARGET — not yet measured
## problem
Manually converting Gherkin/BDD specs into Robot Framework automation blocks was a bottleneck limiting how fast test automation could scale.
## approach
Engineered a LangChain + RAG pipeline with vector-embedded domain knowledge to interpret Gherkin syntax contextually and auto-generate automation blocks, with a Streamlit interface for the team.
## result
0% boost in test automation scalability across automotive OEM programs.
## problem
An automotive OEM's next-gen IVI/cluster platform needed a validated, high-bandwidth Ethernet and audio-bus communication layer with AUTOSAR-compliant configuration.
## approach
Built the Ethernet stack in C++/Java, developed the A2B audio-bus module for low-latency audio transmission, designed a gRPC inter-service layer for modularity, and integrated Vector SilKit for ECU-level simulation and validation.
## result
Successfully demonstrated to the client; earned a Distinguished Contributor Award for the innovation.
CNHi · Senior Developer
## highlights
Rockwell Automation · Firmware Developer
## highlights
personal project
Real-time automotive telematics platform — OBD-II diagnostics, CAN bus decoding, GPS tracking, DTC analysis, driver-behaviour monitoring and fuel analytics behind an async REST API. Published to PyPI and containerised for GHCR with CI/CD.
personal project
Structured Generative AI curriculum for developers — 7 modules and 29 topics (~100 hours) taking LLM fundamentals through to production-grade systems, plus a hands-on Model Context Protocol guide with four build-along projects.
personal project
Full-stack library management application — async FastAPI REST backend over PostgreSQL 16, with a Next.js and Tailwind front end, orchestrated end-to-end through Docker Compose.
planned build
An MCP server that gives Claude, Cursor and other MCP clients automotive diagnostics know-how: DBC decoding, UDS service and negative-response lookup, DTC decoding. Planned for PyPI.
planned build
A terminal-style RAG assistant on this site: ask about projects and articles, get answers that cite their sources. Planned with rate limiting and a hard spend cap.
planned build
A browser-based CAN bus simulator: virtual ECUs, UDS diagnostics over ISO-TP, a decoded frame view and fault injection. Nothing to install.
planned build
A public, clean-room rebuild of AI test automation: natural language or Gherkin to Robot Framework with self-healing locators, benchmarked on an open demo app.
planned build
Upload a candump or ASC log plus a DBC: decodes frames, flags bus-load, timing and heartbeat anomalies, and explains the likely root cause with frame-level evidence.
# certifications & education
PGP – Artificial Intelligence & Machine Learning
University of Texas at Austin
AWS Certified Solutions Architect – Associate
Amazon Web Services
System Design (HLD & LLD)
Scalar Academy
Data Structures & Algorithms
Scalar Academy
Certificate – Project Management
IIM Indore
B.Tech – Electronics & Instrumentation
UPTU, Lucknow
# awards
Distinguished Contributor Award
Ethernet feature innovation in a vehicle-validation product
Golden Team Academy Award
Cross-team collaboration
M-Power Leadership Certification
Team leadership & client communication
Open to embedded/automotive engineering roles, applied AI work, and technical advisory conversations with founders and business owners.
# fastest way to reach me
./send-email.shOpens your mail client addressed to santosh.iimind@gmail.com — or copy the address if you use webmail. You can also find me on GitHub / LinkedIn below.
# or reach me by phone
./call.sh +91-9699678113