About Projects Services Contact Book a Call
Available for work

Senior AI Engineer I build what others pitch.

Voice agents, real-time audio, ML pipelines, and full-stack products — shipped in production, not stuck in a deck. Reachable in seconds.

5+ Years in Production AI
40K+ AI Calls / Day Shipped
60% Voice Latency Reduced

About Me

I'm Bilal Ahmed, a senior engineer who builds AI voice systems, real-time audio pipelines, and ML products that run in production — not prototypes that live in a repo.

I've led product engineering at RTC League, shipped enterprise voice AI at Afiniti, and built low-latency audio at Younite. From DSP and WebRTC to LLM orchestration and AWS-scale deployments.

Need something built? I respond within hours, not days.

40K+ AI voice calls handled daily in production
2.5s → 1s AI agent response latency reduced
40% → 18% Whisper STT word error rate after fine-tuning
MCP + OAuth Published ChatGPT connector for voice platform

My Projects

Real systems I've shipped — not demos, not POCs.

Voice AI Dashboard
40KCalls
1.0sLatency
99.2%Uptime
01

Production AI Voice Platform

Product Lead · RTC League

End-to-end AI voice product: telephony flows, real-time media, LLM orchestration, multi-agent barge-in, and a React/Next.js ops dashboard.

  • ~40,000 calls/day in production
  • 60% latency cut (2.5s → 1s)
  • RAG, denoising, STT-agnostic audio upsampling
Voice AIAWSTwilio
MCP
Claude
ChatGPT
Voice API
OAuth
02

MCP Server & ChatGPT Connector

RTC League · Telecho Platform

Production Model Context Protocol server enabling natural-language control of the voice platform from Claude and ChatGPT.

  • OAuth 2.0 + PKCE authentication
  • Dual transport: stdio & HTTP-SSE on AWS
  • Shipped as a public ChatGPT connector
MCPAgentic AIOAuth
$ whisper --model fine-tuned
Loading model weights...
WER 40% 18%
GPU inference: 4x CUDA balanced
NLU Intent accuracy: 94.7%
03

Enterprise Speech & NLU Pipeline

Software Engineer · Afiniti

Conversational AI stack for enterprise contact centers — STT optimization, intent/entity NLU, and distributed inference at scale.

  • Whisper WER 40% → 18%
  • Multi-GPU load balancer for heterogeneous CUDA
  • MLflow, Triton, and A/B model rollouts
STT / NLPGPU InfraMLflow
For You
AI-powered picks
04

AI Recommendation Engine

RTC League · Saudi E-commerce

Culturally-aware personalization and catalog understanding for a regional marketplace. Custom ranking designed for measurable conversion uplift.

  • Custom ranking & retrieval for regional catalog
  • Designed for measurable CTR / conversion uplift
MLE-commerceRanking
Input
AEC NS VAD AGC
Output
05

Real-Time Audio & VoIP Stack

Afiniti R&D · Younite

Low-latency noise suppression, echo cancellation, and WebRTC-native audio quality for production VoIP and mobile streams.

  • Deep learning + classical DSP (AEC, AGC, DRC)
  • Edge-optimized neural inference
  • WebRTC VAD & jitter adaptation
DSPWebRTCEdge AI

My Stack

Python
PyTorch
FastAPI
React / Next.js
WebRTC
Twilio
AWS
LangChain
Whisper STT
RAG
MCP
Docker
MLflow
DynamoDB
WebSockets
Node.js

What I Build

Voice Bots

Conversational AI agents that handle calls, answer questions, and automate phone workflows with natural speech.

Audio Processing

Transcription, noise removal, speaker diarization, and custom audio analysis pipelines built to your spec.

ML Models

Custom machine learning models trained on your data — classification, prediction, recommendation, and beyond.

Chatbots

Intelligent chat assistants for customer support, lead qualification, onboarding, or internal knowledge bases.

Let's Talk

I respond within hours. Pick whatever works for you.

Book a Call

Pick a date and time — I'll send you a Google Meet link.

MonTueWedThuFriSatSun

Select a date to see available times

Not sure yet? I'll build a free demo.

Describe what you need — I'll build a working proof-of-concept at no cost. You only pay when you love it.

1. Tell me your idea 2. I build a free demo 3. Love it? Then you pay