Challa Ravindranath

Engineer at heart · Strategist at mind

Trust, by design.

Tactics are only worth the strategy they serve.

Career, 2004 to now

  1. Career: 2004 to now. 22 years, 4 spans, 3 countries.
  2. Helsinki, Finland: 2004 to 2014, 10 years. Wipro, then Nokia.
  3. Espoo, Finland: 2014 to 2015, 1 year. Microsoft (ex-Nokia).
  4. Stuttgart, Germany: 2015 to 2017, 2 years. Roche · Siemens.
  5. Hyderabad, India: 2017 to now, 9 years. Microsoft, then home.

trace challa.career · 22y · 4 spans · 3 countries

Each span is a button; focus or hover it to see its attributes.

Focus areas

  1. AI systems

    I build AI agents that can touch real business data without turning into an audit problem.

  2. Distributed systems

    I make a failure traceable from the UI to the back end, whichever team owns it.

  3. Responsible AI

    The guardrails are part of the design, and every AI deliverable goes through Responsible AI review.

Beliefs

  1. 01

    Trust is architectural.

    You can't bolt it on at the end. I put it in the first design, because that's the only place it sticks.

  2. 02

    Correctness has a face.

    In a clinical lab, a failing test is a patient. I've never stopped seeing it that way.

  3. 03

    Prevent the category, not the instance.

    When the same bug came back, I stopped fixing the bug and fixed the design that let it in.

Work

The call. The outcome.

Four calls I made, and what each one changed.

01 · Microsoft · Enterprise AI · 2025 to 2026

Build the agent once. Let it run anywhere.

I set platform and AI direction for a team of fifteen, and designed the agent harness.

The call

I exposed the skills headless, as an API, so the same agent plugs into the tools customers already use. The tool contracts are versioned, and the model only reaches data through them.

The outcome

One integration.We don't rebuild it for each new surface.

Every write validated.Nothing reaches the data without passing the tool contract.

  1. 02 · Microsoft2017 to 2025

    One ID across every boundary.

    Support SLA down 60%. It became the tracing standard across the org.

    The call

    I put one correlation ID across the front-end/back-end boundary where tickets got lost.

  2. 03 · Roche Diagnostics2015 to 2017

    Correctness, on a live clinical line.

    I kept releases stable across four product families, at about 1,400 samples an hour.

    The call

    As system integration lead, I owned the merge of every component into one cobas instrument, and the validation framework behind it.

  3. 04 · Microsoft (ex-Nokia)2014 to 2015

    If trust isn't in the handshake, nothing moves.

    We got carrier certification with Orange, Deutsche Telekom and Vodafone.

    The call

    I validated RCSe at protocol level, SIP signalling against the GSMA RCS specs.

Built in the open

Contributor · Microsoft Agent Framework

AI Playbook

  • Graphs
  • Heaps
  • Hashing
  • Dynamic programming
  • Concurrency

Underneath every AI system are old ideas.

I work through data structures, algorithms and system design the way real AI systems use them.

C# · .NET 10 · tests that run

GitHub

TSight

No model. No guessing.

Point it at a TypeScript codebase and you'll understand it in minutes. Every number checks against the repo. It runs on your machine, so your code stays there.

Mental model · Hotspots · Blast radius · Bus factor · Issues · Codebase map

TypeScript · ts-morph · git history

GitHub

Pulse

The model asks. Config decides.

Ask Azure Data Explorer a question in plain English. Config owns the clusters, schema, auth and filters. The model writes only the filter clause, so it can't invent a cluster or a column.

Python · Azure Data Explorer · YAML · Azure OpenAI · DuckDB

GitHub

Voices

In their words.

Joel KatzevmanMicrosoft

A thoughtful collaborator who uplifts those around him. His combination of technical depth, adaptability, and natural leadership make him an incredible asset to any team.
Principal EM, Microsoft · 5+ years togetherHyderabad · 2017 to 2026

Your visit

Your visit as a trace. Each span links back to its section.

Your path, drawn

corr

One ID across every boundary. Yours, too.

One span is still open: let's talk.

I measured this on your device. Nothing is stored or sent.

Contact

I lead engineers who ship things the business can count on. I'm looking for my next team.

Let's talk.

02 · Microsoft · 2017 to 2025

Tracing that became the standard

The problem

One business operation crossed back-end services, plugins, a data platform and a React front end, with no end-to-end trace. Every team had its own correlation pattern, and support tickets vanished at the front-end/back-end boundary.

The call

I put one correlation ID across that boundary. It sits in the instrumentation layer, so product code never has to carry it. It's thread safe and the type system enforces it. Structured logging picks it up. I did the same thing twice more: a design-token system that cut accessibility regressions, and a CI failure classifier that finally told infrastructure flakes from real regressions.

The value

60%

We cut the support SLA by 60%, and the whole organisation took the approach as its tracing reference.
01 · Microsoft · Enterprise AI · 2025 to 2026

Building the agent harness

The problem

An AI agent over an enterprise platform has to touch real business data, where one wrong write becomes an audit problem.

The call

I gave it one governed path from conversation to model to data:

A memory layer holds conversation state.

A gateway applies guardrails and rate limits on every call.

Models run on a managed, serverless AI platform.

Enterprise data is reachable only through tools.

Skills are exposed headless as an API, so the same agent can plug into the tools customers already use.

Versioned tool contracts mean the model can call tools but never write unvalidated data.

The value

We built the agent once and it runs wherever customers are, with no second integration. Every write is validated, and every AI deliverable clears Responsible AI review. I set direction for a team of fifteen engineers.
03 · Roche Diagnostics · 2015 to 2017

Clinical systems integration

The problem

An instrument like the cobas p 512 is many components built by separate teams: the control unit, sample supply, module sample buffers. They only count once they work as one instrument on a live clinical line.

The call

As system integration lead, I owned the merge:

Brought every component together into one integrated system build: the control unit (system coordination, LIS connectivity), sample supply and module sample buffers.

Did it across four product families: cobas 8100, p 612 / p 512 and p 501 / p 701.

Owned component releases through to the integrated build.

Led the cross-product Python validation framework, including the Python 2 → 3 migration, so it outlived the interpreter.

Later moved to the firmware team, working below the application layer on the instruments themselves.

The value

~1,400

I kept releases stable on live clinical lines across four product families, running about 1,400 samples an hour.
04 · Microsoft (ex-Nokia) · 2014 to 2015

Carrier-grade protocol validation

The problem

Rich Communication Services on Windows Phone could only reach people once each carrier certified it, and certification is decided at protocol level: SIP signalling against the GSMA RCS specifications.

The call

I validated where trust actually gets decided, in the handshake:

Owned RCSe validation at protocol level, against GSMA RCS conformance.

Tests in C#, running in CI.

Logs and SIP traces as the diagnostic surface, so a failure is read in the messages themselves.

The value

3

We got carrier certification with Orange, Deutsche Telekom and Vodafone.