Muhammad A.
AI Voice Agent Developer | Twilio Expert | Conversational AI | Vapi
I build AI voice agents on Twilio, Vapi, and Retell that answer calls, book appointments, and qualify leads, and RAG systems that turn large document libraries into instant, accurate answers. Expert-Vetted (top 1% on Upwork), Top Rated Plus, $300K+ earned across 250+ projects and 9,900+ hours. If you need an AI receptionist that handles inbound calls 24/7, an outbound voice agent for lead qualification and follow-ups, or an LLM assistant grounded in your own documents, this is the exact work I do every day. WHAT I BUILD AI Voice Agents (Twilio, Vapi, Retell AI, Deepgram, ElevenLabs) ➤ Inbound AI receptionists: call answering, FAQs, appointment booking, smart transfer to a human when needed ➤ Outbound agents: lead qualification, appointment reminders, follow-up calls ➤ Real-time call transcription, sentiment analysis, voicemail detection, and call summaries pushed to your CRM ➤ Twilio end to end: number porting (including from Grasshopper), A2P 10DLC registration, SIP trunking, Studio flows, WhatsApp Business, call routing and recording, SMS automations, PBX extension transfers RAG and LLM Systems (AWS Bedrock, Claude, GPT, LangChain, pgvector) ➤ Retrieval augmented generation over large corpora: chunking, embeddings, vector search, reranking ➤ Document Q&A assistants and AI chatbots that cite sources instead of hallucinating ➤ Ingestion pipelines with OCR (Textract), queues (SQS), and cost monitoring Full-Stack Delivery Node.js, NestJS, FastAPI, Python, Next.js, React Native, PostgreSQL, MySQL, Redis, Docker, AWS (EC2, RDS, Bedrock, SQS). I ship the voice AI agent, the automation around it, and the product it lives in: dashboards, auth, billing, analytics. SELECTED WORK 🟢 SlayAI (legal AI platform): architected the RAG pipeline serving ~4,000 active users over a 300GB+ legal document corpus on AWS Bedrock, pgvector, SQS, and Textract. 97% answer accuracy in internal evals, with automated document ingestion and readiness checks. 🟢 HopDial (VoIP + AI calling platform): built live-call AI transcription and sentiment analysis on Twilio infrastructure with Deepgram STT and ElevenLabs TTS. 🟢 WhoseNo (my own product): high-traffic phone intelligence platform with 5.79M+ records, MySQL and Redis at scale behind Cloudflare, plus a companion Android app. WHY CLIENTS KEEP ME 100% Job Success across 180+ reviews. Most of my work turns long-term: clients bring me in for one agent, and I end up owning their AI roadmap. I communicate daily, demo early, and never disappear. HOW TO START Send an invite or book a consultation. Tell me your call volume or document volume and the outcome you want. I'll come back with an architecture, timeline, and fixed scope.
Earnings and projects
Last 6 months ÷ 6
Last 6 months: $36k
Billed in all: $394,486$394k
$3.3k in the last 6 months
Across 327 contracts
Earnings, last 30 days
Skill rankings: Top 1% in ElevenLabs
Ranked by MRR among everyone on Upwork with the skill.
- Vapi140 freelancers · 🇵🇰 #2Top 2%🇵🇰#2 of 58
- #4Retell AI116 freelancers · 🇵🇰 #3Top 3%🇵🇰#3 of 58
- #5Amazon Bedrock204 freelancers · 🇵🇰 #2Top 2%🇵🇰#2 of 44
- #7ElevenLabs638 freelancers · 🇵🇰 #3Top 1%🇵🇰#3 of 176
- #7AI Speech-to-Text336 freelancers · 🇵🇰 #3Top 2%🇵🇰#3 of 82
- #8AI Text-to-Speech529 freelancers · 🇵🇰 #3Top 2%🇵🇰#3 of 122
- #14Twilio743 freelancers · 🇵🇰 #4Top 2%🇵🇰#4 of 257
- #16AI Voice Agent Development429 freelancers · 🇵🇰 #10Top 4%🇵🇰#10 of 192