Preetam Ramdhave
Forward Deployed Engineer · Seattle, WA
The work speaks.
I realized something early: the best software in the world doesn't matter if it doesn't fit where the customer actually lives. The constraints, the legacy systems, the organizational politics, the user who is tired and just wants to get through their day. You can't spec your way to that understanding. You have to be there.
Where it started
It started at KPIT in 2008. I was 25, shipped to UK enterprise clients as the single point of contact between an offshore engineering team and a customer who had very specific, very real problems. No PM buffer. No requirements handed down from above. You sat with the customer, you learned how they worked, and you built for their reality.
Eight years. Three "Delighted Customer" awards voted by the clients themselves. I didn't know it had a name yet — Forward Deployed Engineering — but the motion was already set: embed, discover, design, ship, operate, generalize.
The evolution
In 2018 I moved to Seattle and started applying the same motion to AI — first as an internal FDE at a Fortune 500 industrial manufacturer, then as a founder shipping production products across ed-tech, healthcare, and spiritual-tech.
The tools changed. The motion didn't. The customer is still in the room. The problem is still never the stated problem. The work still has to survive contact with production.
Proof moment
60–80% effort reduction. First-of-kind enterprise agentic AI.
A Fortune 500 industrial manufacturer needed to automate compliance and policy checks across hundreds of high-stakes, 200+ page customer documents. Manual review was slow, inconsistent, and highly prone to audit failure. Leadership knew AI was the answer, but the constraints were massive: no external data transmission, deterministic output, full auditability to specific pages and standards, and no black-box failures.
I architected and deployed the organization's first production agentic AI workflow. The solution used an event-driven pipeline—API Gateway to Step Functions orchestration, parallel Lambda tools for chunk planning and policy validation, and RAG grounding on an AWS Kendra GenAI Index. By decomposing the architecture into modular, testable stages with strict JSON schema validation and coverage gating, we delivered a system that guaranteed 100% review coverage while eliminating the silent partial failures typical of LLMs.
Production systems shipped end-to-end
Manual effort via first-of-kind agentic AI workflow
Processed in 48 hours at a healthcare event
Delighted Customer, voted by UK clients
Why this work
Most software is built for a spec. FDE work is built for a reality. The customer's reality — their legacy systems, their team's actual skill level, their budget, their deadline, their ambiguity. That gap between spec and reality is where most AI deployments fail. It's also where I do my best work.
I don't build demos. I build systems that customers depend on when the event is tomorrow and the stakes are real. That requires a different kind of ownership — not just "I shipped the code" but "I was there when it ran in production, and I fixed what broke."
The motion
Set at KPIT with UK clients. Refined across Fortune 500 and solo founder products.
Experience
Principal AI Architect · Enterprise Systems Lead
Fortune 500 Industrial Manufacturing Enterprise
Renton, WA
Current
- ▸Designed and shipped the organization's first production agentic AI workflow — event-driven ingestion via API Gateway → Step Functions, parallel Lambda tools, RAG grounding on AWS Kendra GenAI Index. Outcome: 60–80% reduction in manual document review effort.
- ▸Architected a cross-account S3 secure file distribution platform with AWS Transfer Family (SFTP), KMS encryption, and home-directory isolation. Authored reusable vendor onboarding documentation now standard across the team.
- ▸Led legacy application modernization: defined cloud-native migration path with phased roadmap, produced formal TIDs and ADRs with board-ready architecture reports.
- ▸Built AWS-to-Azure migration frameworks mapping Lambda/S3/Step Functions to Azure equivalents for org-wide multi-cloud strategy.
Full-Stack & AI-Native Product Development · External Client Engagements
Independent Founder & Embedded Engineer
Seattle, WA
2018 — Present
- ▸OmmSai: LLM document pipeline processing 15,000+ handwritten prescription PDFs in 48 hours for a charitable healthcare event. Claude Sonnet + Google Drive API + ThreadPoolExecutor + Tkinter GUI. Open-sourced.
- ▸ScholarPath: Active production ed-tech platform for Maharashtra MSCE scholarship exam prep. React + TypeScript + FastAPI + Supabase + Razorpay. Parent-as-gateway model, 1,000+ students.
- ▸JapaApp: Spiritual mantra-tracking PWA live in production. Originally on AWS (Lambda, RDS Proxy, Cognito, SAM); owned the migration decision to Firebase. Razorpay tiered donation flow.
- ▸Trading System: Automated IBKR futures trading with vertical spread options, NLP command parsing, React dashboard.
Technical Lead / Sr. Software Engineer · Embedded Engineer for UK Enterprise Clients
KPIT Technologies (formerly KPIT Cummins Infosystems)
Pune, India · Onsite UK engagements
Jun 2008 — 2017
- ▸Eight years as single point of contact for UK enterprise clients — capturing requirements onsite, designing systems for their reality, owning analysis through deployment.
- ▸Three consecutive "Delighted Customer" awards (2012, 2013, 2014) voted by the client for direct impact.
- ▸Real-Time Device Communication Platform: Scalable UDP socket server for concurrent VPS security device communication, WCF services for CRC-based authorization, video/image extraction from raw byte streams. .NET 4.0, C#.
- ▸Order-to-Invoice Enterprise Platform: Full PRCR cycle ownership — analysis, design, implementation, regression, release. Complex contract handling and bespoke invoicing calculations. .NET 2.0, VB.NET, SSRS, NHibernate, SQL Server.
Core Competencies
AI / LLM
Cloud & Infra
Backend
Frontend
Security
FDE Practice
Education & Certifications
Languages: English · Hindi · Marathi
What I'm looking for next
Technical leadership and enterprise architecture roles at top-tier tech firms, AI labs, and high-growth scaleups — organizations building complex, resilient systems where the question isn't "should we use AI" but "how do we architect it to actually work at scale."
Best fit
Technical leadership · Enterprise architecture · Core AI platforms
Open to
Principal roles · Staff engineering · Technical advisor
Not exploring
Full-time at non-AI companies
Based in Renton, WA — in the shadow of Boeing and a short drive from Amazon, Microsoft, and the rest of the Pacific Northwest tech corridor. Grew up in Maharashtra, India; still use Marathi phrases in internal docs as a nod to where it started. I practice japa — daily mantra repetition, which is partly why I built JapaApp. आपलं काम बोलतं — the work speaks.