Senior Software Engineer & Product Builder
Himanshu Sharma builds production SaaS end to end in Java, Spring Boot, AWS, React, and applied AI.
I independently designed, built, and am commercializing PeakInsight, a multi-tenant B2B SaaS for travel agencies. It covers the CRM, itinerary authoring, payments, and AI workflows that run on WhatsApp, voice, and PDFs. I own it end to end: product decisions, backend architecture, frontend, AWS infrastructure, and customer demos.
Looking for senior engineering roles at remote-first SaaS and AI product companies. [TODO: location / time-zone overlap]
Models propose, services commit
LLM output is structured JSON that backend code resolves against real catalog and CRM records. The voice workflow returns a draft; only existing, permission-checked APIs write.
Queue what you don't control
Third-party webhooks are signature-checked, acknowledged, and moved onto SQS FIFO queues, then processed by workers with a fixed in-flight limit.
Make cost and failure visible
Every AI provider call is recorded as a usage hit and priced into credits. Logs are structured JSON carrying request, user, and agency IDs.
Flagship project
PeakInsight: an AI-native travel operating system
Travel agencies run sales across WhatsApp threads, supplier PDFs, phone calls, and spreadsheets. PeakInsight puts the lead → trip → package → payment loop in one multi-tenant system and builds AI into those workflows, metered like any other billable resource.
- What I built: the Spring Boot backend (3 Maven modules), React web app, AWS setup, every AI workflow, and the external integrations.
- Hardest parts: making webhook processing safe to retry, grounding LLM output in each agency's own data, and metering AI usage per tenant.
- Status: in production and being sold to agencies. A public demo runs without signup.
Engineering highlights
Four systems worth a closer look
Each links to a deep dive covering the problem, design, edge cases, trade-offs, and what I'd change next.
Voice → itinerary drafts →
Multi-clip speech-to-text, two-stage Claude extraction, and backend entity resolution produce a preview-only package draft with capped clarification rounds.
Event-driven WhatsApp assistant →
Meta webhooks → SQS FIFO with per-conversation ordering → bounded workers → an intent-routed orchestrator grounded in the agency's own packages and hotels.
AI usage metering & subscriptions →
Per-call token and audio-duration metering, tariff-based credit pricing, two-pool wallets, and a daily subscription lifecycle that suspends tenant access on expiry.
Automations runtime →
Trigger-matched enrollments with ACTION and WAIT steps, a resume scheduler, per-step audit records, and exits when the customer replies or an agent takes over.
Technical expertise
What I work with
Everything listed is used in production code I wrote for PeakInsight. [TODO: add skills from previous roles that PeakInsight doesn't show]
- Backend
- Java 21Spring Boot 3JWT auth (RSA)JPA / HibernateMySQLMaven multi-module
- AWS
- SQS (FIFO)S3RDSSecrets ManagerKMSSESElastic BeanstalkLambdaCloudWatch / SNS
- Frontend & mobile
- React 18TypeScriptViteTanStack QueryRedux ToolkitFlutter
- Applied AI
- Anthropic Claude APIStructured JSON extractionSpeech-to-text pipelinesLLM usage meteringMulti-stage prompting
- Integrations
- WhatsApp Cloud APIMeta Lead Ads webhooksGmail APIIMAP / SMTPFirebase Cloud Messaging
Professional background
Experience
PeakInsight · Founder & Engineer
[TODO: dates]- Designed and built a multi-tenant B2B SaaS for travel agencies: Spring Boot backend, React web app, and AWS infrastructure.
- Built a voice-to-itinerary pipeline, a WhatsApp AI assistant on SQS FIFO, AI usage metering with subscription billing, and an automations runtime.
- Ran product demos with travel agencies. Their feedback shaped the core constraints: preview before save, grounding AI in the agency's own data, and human handoff.
[TODO: Previous company · Title]
[TODO: dates][TODO: 2–3 bullets: systems owned, scale you can verify, team size, stack]
Education
[TODO: degree, institution]
Let's talk
Open to senior backend or full-stack roles at remote SaaS and AI product companies. Happy to walk through PeakInsight's code and architecture in an interview.