Vedant Joshi

I build and ship AI products end to end — currently running Valkoma AI, a B2B SaaS platform with five live modules and paying customers, built solo across product, LLM architecture, and go-to-market. Alongside that, I work on AI data annotation and RLHF projects with global dataservices providers, which keeps me close to how AI systems are actually trained and evaluated. That combination — building products and working inside the data pipeline that feeds them — is what I bring to operational and program roles in AI.

Shortlisted, Google for Startups Immersion Program (Phase 1).

Founder & AI Product Manager at Valkoma AI (2024-01 – Present)

• Built a 5-module B2B AI SaaS platform solo — chatbot builder, website builder, business auditor, AI consultant, competitor analyzer — from an empty repo to paying customers, owning product, engineering, and GTM.
• Ran 15+ customer discovery calls across 4 SMB segments before writing a line of code; the resulting product changes lifted task completion by 35%.
• Wired together a multi-LLM backend (OpenAI, Gemini, OpenRouter) with RAG retrieval and n8n/Make automations on Firebase + GCP — this is what cut client manual work by 60%, not a slide deck.
• Sole point of contact for engineering decisions, sales conversations, and customer support in the early months; later split feature cycles into weekly sprints, which brought delivery time down 40%.
• Selected for Google for Startups Immersion Program, Phase 1 — a national shortlist, not a self-nomination.

AI Data Specialist (Annotation & QA) at RWS Group (2025-09 – Present)

• Work on LLM training-data projects, applying structured validation and quality checks across annotation tasks — current focus is comparison and review work where consistency and accuracy directly affect model output quality.
• Built the habit of cross-checking flagged outputs against guideline criteria before sign-off, since one missed inconsistency early in a batch compounds across the full queue.

AI Data Specialist at Innodata (2025-11 – 2026-01)

• Set quality standards across annotation projects and brought error rates down 25% by building structured validation checks instead of relying on spot-checks.
• Kept finding the same handful of mistakes recurring across different batches — built a simple classification system to flag and group them by type, which is most of what cut the error rate.

Operations Analyst at Zepto Dark Store (2024-10 – 2025-01)

• Tracked daily fulfillment KPIs for a high-volume dark store; the gaps I saw there directly shaped how I later thought about AI product design.

Project Management Associate at Excelerate (2025-12 – 2026-02)

• Ran sprint planning across several client workstreams at once; tightened backlog prioritization enough to move team velocity in a measurable way.

Logistics Supervisor at Schnellecke Jeena Logistics (2024-02 – 2024-06)

Supervised logistics coordination and identified routing inefficiencies adding weekly cost.

Outreach & Program Coordination at TheMentor (2021-09 – 2022-02)

Ran outreach for an education program, working with institutional partners on enrollment.

B.Tech in Mechanical Engineering – Bajaj Institute of Technology, Wardha (2018 – 2022)

Program in Public Policy & Political Consulting – Rebounce Public Policy & Political Consulting (2026)