Google Student Ambassador B.Tech CSE (Cyber Security), SCSVMV University

Gopal Tenneti

I build software that stays correct under load: multi-tenant SaaS, data systems and AI pipelines, shipped all the way to production.

Selected work

Two systems I designed and built end to end. One runs a real business workflow in production; the other checks its own answers.

Align

Live in production

A multi-tenant sales CRM for small teams.

Leads, contacts, companies, deal pipelines, proposals, activities and tasks, with dashboards, sales reports and subscription billing across three plans. The hard part was not the screens. It was making sure one company can never see another company's data, and that every multi-step action either fully happens or doesn't happen at all.

  • Isolation in the database itself. PostgreSQL Row-Level Security enforces tenants and three roles, so a bug in the app can't leak data.
  • All-or-nothing workflows. Lead conversion, bulk edits and resumable CSV/Excel imports of up to 10,000 rows run as single transactions.
  • Background automation. Tracked proposal emails, follow-up reminders and two-way Google Calendar sync run on Edge Functions and scheduled jobs.
  • Released like a real product. Staging first, backups before every release, versioned migrations and 250+ automated tests.
How a request moves through Align
React appTypeScript, Vite
PostgreSQL on SupabaseRow-Level Security checks every row, per tenant and role
Edge FunctionsLogin security, emails, calendar sync
Scheduled jobsFollow-ups, billing checks, import cleanup
Razorpay Resend Google Calendar
How AdaptIQ answers a question
  1. Question comes in
  2. Meaning searchHuggingFace embeddings in Pinecone Keyword searchBM25
  3. Merge the two rankingsReciprocal Rank Fusion
  4. Answer, then grade itGemini 2.5 Flash as judge: faithfulness and context precision
  5. Low score? Re-routeAutomatic fallback, every step logged to Neon PostgreSQL

AdaptIQ

A question-answering system that grades its own answers.

Most RAG demos return an answer and hope it's right. AdaptIQ searches two ways at once, merges the results, then has a second model score the answer before anyone sees it. When the score drops, it switches strategy instead of guessing.

  • Hybrid retrieval. Dense semantic search and BM25 keyword search, combined with Reciprocal Rank Fusion.
  • LLM-as-a-Judge. Gemini 2.5 Flash scores faithfulness and context precision on every answer.
  • Measured, not guessed. Full telemetry in Neon PostgreSQL for tracking quality across sessions.

Credentials

Recognition and certifications

  • Google Student Ambassador

    Google, selected May 2026

    I represent Google's developer ecosystem on campus: running sessions on Google Cloud and AI tools and growing a community of student builders.

  • Google Arcade Program, Cloud Skills Boost

    Google Cloud skill badges

    Hands-on, lab-graded badges in Google Cloud infrastructure, AI services and developer tooling.

Education

  1. B.Tech, Computer Science and Engineering (Cyber Security)

    SCSVMV University, Kancheepuram, 2024 to 2028

    8.10 CGPA overall, 8.45 in semester 4

  2. Class XII, Karnataka State Board

    Science with Computer Science, 2024

    89.16%

  3. Class X, Karnataka State Board

    2022

    95.84% and first rank in school

Certificates

Verify on GitHub

    Toolbox

    Strongest in Java and Python, PostgreSQL, concurrency and system design.

    Open to internships.

    Backend, platform or AI engineering. Based in Kancheepuram, Tamil Nadu, happy to work remotely or relocate.

    tennetigopal04@gmail.com