SujayVittal .
Engineering Manager, AI at Samsara. Helping teams ship AI systems that earn trust — at the Go-To-Market edge of the physical economy.
About
“Building tools that support people behind the scenes.”
— Alumni Spotlight, Purdue University Northwest
I'm an engineering manager in Seattle leading AI engineering for Samsara's Go-To-Market engine. Before that, I spent six years at Microsoft and Microsoft AI shipping generative AI into products millions rely on — from Copilot Search to the MAIA inference platform.
My work sits at the intersection of research and reliability — bridging model experimentation with the scalable infrastructure it takes to put AI in front of real users. I care about thoughtful design, strong processes, and the slow craft of building software that earns trust.
Outside of work, I'm logging flight hours as a student pilot, making pour-overs on the kitchen counter, arguing about cricket, and chasing the right record for the right mood.
6+
Years shipping AI systems
2020–2026
At Microsoft AI
2
Publications
↗︎
Flight hours, logging
A decade of shipping systems people rely on.
From a student-led startup in Bengaluru to architecting Copilot Search at Microsoft AI to leading GTM AI engineering at Samsara today.
Feb 2026 — Present
Seattle, WA
- Leading AI Inference and Automation engineering teams inside the Revenue Operations AI & Data org — building the future of the Go-To-Market engine.
- Shipping production-grade Generative AI: RAG systems and LLM copilots that fold real-time insights and sales intelligence directly into the GTM funnel.
- Bridging model experimentation and scalable inference infrastructure so Samsara can engage the global physical economy through personalized automation.
Mar 2020 — Jan 2026
Redmond, WA
- Architected and delivered the end-to-end Orchestrator for Copilot Search — dynamic query analysis and classification routing requests to LLM pipelines with MAI DeepSeek for context-aware generation.
- Primary SWE for high-impact work: Responsible AI Classifier plugin for Copilot Search and migration of generative search components from DLIS to XAP.
- Spearheaded AI copilot initiatives and stood up an AI lab inside Core Financial Services; built a separate Azure OpenAI solution to streamline Azure DevOps pull requests and fine-tune models.
- Primary SWE / lead on transitioning MAIA's distributed inference services from Service Fabric to Kubernetes; led stress testing for the MAIA Control Plane and optimization of Azure OAI inference workloads.
Aug 2016 — Dec 2017
Bengaluru, India
- Optimized search algorithms for the BusinessObjects platform — back-end and front-end work that reduced unwanted traffic and improved system performance.
2013 — 2015
Bengaluru, India
- Led Google's student ambassador initiatives across Bengaluru — organizing developer meet-ups, campus workshops, and tech-community events that connected students to Google's products and engineering culture.
- Built and ran peer learning programs for computer science students; grew an active on-campus developer community and ran launch events for Google tools and platforms.
Aug 2012 — May 2015
Bengaluru, India
Stealth
—Co-Founder & CTO
- Co-founded an early-stage venture out of college; led engineering and product for a social-messaging product from zero to a live user base.
- Built and led a volunteer engineering team; recognized at regional pitch and startup competitions.
Languages
- Python
- TypeScript
- Go
- C#
- Java
AI / ML
- TensorFlow
- Azure OpenAI
- AzureML
- RAG
- LLM Orchestration
- Fine-tuning
Infra
- Kubernetes
- Service Fabric
- Azure
- GCP
Web
- Next.js
- Node.js
- AngularJS
- .NET
Projects that try to earn their complexity.
A web-OS for personal context. A model of economic inequality. A map that works without GPS. Different years, same instinct — build tools that are useful in the real world.
Intelligent Inflation & Income Prediction
Characterizing wealth disparity with Wide & Deep learning.
Research at Purdue University Northwest assisting a professor studying economic inequality in the United States. Led a team of six undergrads and grads building robust metrics and predictive models for inflation and income.
Disaster Rehabilitation Modeling
Cloud NLP applied to crisis response.
Built an effective model for disaster rehabilitation using GCP's Cloud Natural Language API. An extension of "Empirical Analysis of Social Networks," published at the International Conference on Computer Science and Technology.
CompMaps
Offline navigation from steps and compass.
A prototype map that works without GPS — navigating by pedometer steps and magnetic heading alone. Adopted at school for internal navigation.
Four things I’d talk about for hours.
The parts of the week that make the engineering better. Move your cursor across.
Coffee◐
Espresso mornings on the Rocket — single-origin, 9-bar, scale on the drip. Filter coffee at dusk — dark roast, chicory, dabara, jaggery.
Open↗︎01/04Cricket✦
Grew up on the late cuts of Dravid, the swagger of Sehwag, and late-night Ashes radio. Still keep a bat within reach.
02/04Music♪
Writing, recording, and releasing original music — a second language for the parts of life engineering can't capture. New singles out on streaming.
03/04Flying↗︎
Weekends at the pattern — VFR, crosswind landings, and learning to trust instruments. The sky clears the mind.
04/04Currently brewing
Ethiopian Yirgacheffe
Rocket Appartamento · 9 bar · 18 → 36 g
Currently reading
Stick & Rudder
Currently watching
India tour · any format
Papers, podiums, and a proud alumni feature.
International Conference on Computer Science and Technology
Empirical Analysis of Social Networks
ISBN: 978-93-85225-62-8 — co-authored peer-reviewed publication on social network analysis.
Howard University — Social Media Technology Conference
Oral presentation: disaster rehabilitation
Presented the disaster rehabilitation solution in Washington, DC.
06 · Let's talk
Say hello.
For interesting engineering problems, AI system deep-dives, or simply to compare flat whites. I read everything.