One engineer. Four products. All in production.
A solo builder studio shipping commercial-grade astronomy, applied-AI, and observability software — money-correct payments and physics-correct math. No funding. No padding.
| Product | What it is | Status |
|---|---|---|
| Kriya | Commercial astronomy API. 129 endpoints across Western, Vedic, Hellenistic, Jaimini, KP, and electional traditions. First-principles VSOP87D + ELP/MPP02 + DOPRI8 engine, zero third-party astronomy libraries. 1,398 Horizons + Meeus-validated tests. | engine v9.x · live at kriya.insightsbyomkar.com |
| Urja | Generative-visual API and Critic. Cross-provider taste-auditor: the judging LLM is enforced in code to be a different provider than the generating LLM. | v0.9 · live at urja.insightsbyomkar.com |
| Netra | Browser-native astrology workbench. BFF pattern composing Kriya endpoints. Ships zero astronomy of its own. | v2.32+ · live at netra.insightsbyomkar.com |
| Bodha | Homegrown observability library — error tracking, structured event schema, SDK. Library, not a service. | v0.2 · internal |
Plus the consumer app at insightsbyomkar.com, backed by a 27,000+ member community. Multi-tier AI support agents, dual provider rails (OpenAI + Anthropic), idempotent Stripe payments with an atomic credit ledger, premium PDF reports, and a programmatic SEO library. Live, early traction.
~4,500 automated tests across the platform — real suites, not stubs.
One engineer. The full stack — system design, backend, applied AI, ops, billing, growth.
- Engine work is first-principles. No third-party astronomy library — just JPL ephemerides, Meeus, and tests. Validated to sub-arcsecond against JPL Horizons.
- Payments are correct under concurrency — idempotent webhooks, row-locked atomic credit ledger, delta-based refunds, automatic refund-on-failure, a two-person approval state machine, and a drift-detecting refund executor.
- AI architecture is multi-tier — RAG, multi-agent support, and a cross-provider LLM-judge that won't let a model grade its own provider's output.
- Discipline is versioned releases, real CHANGELOGs, RLS on every user-scoped table, and no fake "team" voice.
- Kriya astronomy API — case study · architecture, accuracy disclosure, source policy
- Solo-shipped consumer SaaS — case study · product strategy, payments, AI agents, operating rhythm
- Kriya SDKs · three public clients — TypeScript (npm) · Python (PyPI) · Go — zero-dependency
Omkar Jaliparthi · Founder & Principal Engineer · Backend & Applied-AI Engineer (7+ yrs) · ex-Tech Mahindra, Capgemini · MS CS · San Jose, CA.
Open to Founding Engineer · Senior Backend · Forward-Deployed / Applied-AI roles. Will relocate US-wide. Authorized to work in the US (EAD) · no sponsorship required.
Numbers, not vibes. See the changelogs.