Vamsi Alla

Forward Deployed Engineer · Hospitality AI

I build AI that works at the front desk, not just in a demo.

I take messy operational problems, turn them into AI systems, measure whether they're right, and hand them over so teams can run them. My focus is hospitality: hotels, guests, and the staff who serve them.

See the hotel assistant GitHub Resume (PDF)

Beach Resort · "A Gold member wants late checkout on Saturday. How late can they stay?"

On Saturday, a Gold member's late check-out is capped at 12:00 PM for all guests, including Gold members1. The brand-wide Gold benefit of 2:00 PM does not apply on weekends because the Beach Resort property rule overrides it14.

1 Beach Resort policies › Late Check-out (Property Override)   4 Guest Rewards Loyalty Program › Tiers and Benefits

Work

The live Hotel Operations Knowledge Assistant: a front-desk agent clicks a suggested question about dogs at Beach Resort and a cited answer streams in
▶ Watch the demo Try the live app

Hotel Operations Knowledge Assistant

Front-desk staff ask policy questions in plain English and get a short answer with the source for every fact, for the right hotel. It refuses when the documents don't say, corrects wrong facts in the question, and points to a manager for refunds and safety. Managers see what staff ask and where the policies have gaps.

  • Python
  • Claude Haiku 5.5
  • Hybrid search (BM25 + embeddings)
  • Streamlit
  • pytest
  • GitHub Actions
  • Docker
Case study Live demo 2-minute video Code
96–98%
answer accuracy on a 54-question eval, incl. prompt injection and must-refuse cases
$0.17
measured cost per 1,000 questions, about 2.5 s per full answer
274
tests, 97% coverage, dependency audit and eval gates in CI on every commit
~$16K
projected yearly value per property (assumptions)

In progress

Guest Request Agent

Reads guest emails, checks bookings and policy, drafts replies, and waits for staff approval before sending.

Planned

AI Evaluation Harness

100+ labeled cases, hallucination checks, and regression gates that block bad changes before release.

Planned

Production Deployment

Latency, cost and privacy limits: model routing, personal-data redaction, and a monitoring dashboard.

Planned

Client Handoff Package

Architecture docs, decision records, incident runbooks, and a recorded walkthrough for the client's team.

How I work

Every project follows the same path as a real client engagement.

  1. UnderstandFind the real problem and what it costs the business.
  2. BuildWork with the data as it really is, messy and inconsistent.
  3. MeasureProve quality with evals, not impressions.
  4. DeployShip under real limits: speed, cost, privacy.
  5. Hand offDocs and runbooks so the team can own it.

About

I'm Vamsi Alla, an engineer focused on bringing AI into hospitality operations.

I'm looking for Forward Deployed Engineer and AI engineering roles where I can work directly with customers to solve real operational problems.

Contact

Open to Forward Deployed Engineer and AI engineering roles.

vamsialla.dev@gmail.com

LinkedIn

GitHub