Surya Teja Reddy Dwarampudi

• Build training environments and evaluation infrastructure as part of the founding engineering team.
• Work with 3+ frontier AI labs across the world on model evaluation across more than 5 workstreams.
• Develop data pipelines and internal research tools; mentor junior engineers across the team.

• Built sales and operations platforms, including an in-house CRM that replaced LeadSquared with 100% adoption in 4 months.
• Owned frontend architecture, Elasticsearch search, role-based access control across 12+ modules, and a React Flow workflow builder.
• Developed voice agents, automated call auditing, and an AI simulator for sales training. Project details and results are below.

• Needle-Mover Award: reduced mentor no-shows from 11% → 3% via a custom scheduling feature on Ruby on Rails + React.
• Self-serve admin tools cut weekly support tickets by 60% (30→12), saving 100+ eng-hours/yr.
• Improved the Article page for 500K+ MAU using lazy loading and server-side pagination.

• Built interactive admin dashboards in Django to visualize inter-school activity data.

Built a calling platform that handles sales conversations and books meetings. Combined Gemini, Deepgram speech recognition, and ElevenLabs speech synthesis, with Redis and WebRTC for call orchestration.

Stack: Gemini, Deepgram, ElevenLabs, Node.js, Redis, WebRTC

Role: Lead engineer · architecture, latency, orchestration

The task

Handle sales conversations and meeting booking through a real-time voice agent.

What I built

Led the architecture, latency work, and call orchestration. Combined Deepgram speech recognition, Gemini, and ElevenLabs speech synthesis with Redis and WebRTC.

The result

Average latency below 800 ms, more than 15 concurrent calls, and round-the-clock operation.

• <800 ms — average latency
• 15+ — concurrent calls
• 24/7 — operation

Replaced LeadSquared with an internal CRM, cutting annual costs by 90%. Built lead filtering with sub-millisecond search latency across 3M+ leads and 100K+ monthly queries, role-based permissions, a workflow builder, and support for concurrent A/B tests.

Stack: Next.js, Turborepo, Elasticsearch, PostgreSQL, React Flow, Kafka

Role: Frontend architecture · search infra · RBAC · workflow engine

The task

Replace LeadSquared with an internal CRM for the sales team.

What I built

Owned frontend architecture, search infrastructure, role-based permissions, and the workflow engine. Built filtering over 3M+ leads, permissions across 12+ modules, a React Flow workflow builder, and support for concurrent A/B tests.

The result

Reduced annual costs by 90%, saving ₹5.4 Cr+ per year. Reached full team adoption in four months. Search served 100K+ monthly queries with sub-millisecond latency.

• 90% — cost reduction
• 3M+ — leads indexed
• 100% — team adoption

Built a pipeline that transcribes sales and support calls, checks them against compliance policies, and sends flagged passages to reviewers. Added review tooling and coaching feedback, with 94% precision on flagged issues.

Stack: Python, FastAPI, OpenAI, AWS MSK, Postgres, OpenSearch

Role: Pipeline architecture · LLM prompting · review tooling

The task

Check sales and support conversations against compliance policies and surface passages for human review.

What I built

Built the transcription and audit pipeline, LLM prompting, review tooling, and coaching feedback screens.

The result

Processed 10K+ calls per month with 94% precision on flagged issues and a three-day review-to-production loop.

• 10K+ — calls/month
• 94% — precision on flagged issues
• 3 d — review-to-prod loop

Built a training platform where sales reps practice conversations with AI prospects. Created a library of 40+ personas, an evaluation harness, and feedback screens. The platform increased practice volume fivefold and cut ramp-up time by six weeks.

Stack: Gemini, TypeScript, Next.js, Postgres, pgvector

Role: End-to-end · prompts, eval harness, frontend

The task

Give sales reps a place to practice conversations with AI prospects.

What I built

Built the platform end to end: persona prompts, an evaluation harness, and the frontend for practice and feedback. Created a library of more than 40 personas.

The result

Practice volume increased fivefold and rep ramp-up time fell by six weeks.

• 6 wks — rep ramp-up cut
• 40+ — persona library
• 5x — practice volume

Build GUI environments in which single and multi-agent systems can perform desktop tasks. Author SWE-Bench and Terminal-Bench evaluations, process task traces into training data, and develop tools for the research team.

Stack: Python, TypeScript, Playwright, Docker, RL · PPO/GRPO

Role: Software Engineer II · environments, evaluation, research tooling

The work

Build GUI environments in which single and multi-agent systems can perform desktop tasks.

My contribution

Author SWE-Bench and Terminal-Bench evaluations, process task traces into training data, and develop internal research tools. I also mentor junior engineers.

Team context

Part of the founding team at Evaratus AI, formerly Scaler AI Labs. The organization has more than 50 engineers, works with 3+ frontier AI labs across the world, and spans more than 5 workstreams.

• 50+ — engineers in org
• 3+ — frontier AI labs
• >5 — workstreams