Co-Founder, Sorted Trader Ltd · AI Engineer · MSc Data Science

I co-founded Sorted Trader Ltd to address the operational and regulatory complexity of the UK used-car market. The platform consolidates dealer operations — stock, compliance, payments and handover — while giving buyers verified vehicle information and a transparent purchase process.

📫 Email · LinkedIn

Buying and selling a used car in the UK is slow, paperwork-heavy, and regulated at nearly every step — FCA permissions, Consumer Duty disclosures, AML checks, pre-delivery inspections, CRA windows. Dealers stitch this together across half a dozen disconnected tools. Buyers, meanwhile, get almost none of it surfaced to them.

Sorted Trader collapses that into one platform, with an agent-accessible interface layered on top so the whole thing can be driven conversationally.

For dealers

• Stock management, vehicle lookup, MOT history, photos, and prep-pipeline tracking
• Appointment scheduling, deposits, payments, and Stripe-backed settlement
• Compliance built into the workflow — FCA, AML, CRA, PDI, pre-contract disclosures
• Financial OS: true cost per unit, recon spend, days in stock, VAT return export, cashflow
• Buyer Vehicle Reports and generated sale and handover packs

For buyers

• Search and compare vehicles, then book appointments with the dealer directly
• ULEZ compliance and outstanding recall checks before you commit
• Total cost of ownership estimates, deposits, and finance pre-qualification
• Digital handover pack, post-purchase issue reporting, and GDPR self-serve data export

Under the hood — an MCP (Model Context Protocol) server exposing the platform as scoped tools, so both dealer staff and buyers can operate it through an AI assistant rather than a dashboard. Scopes are enforced server-side; the tool surface changes based on who's authenticated.

• Field Summary AI Engine — FastAPI microservice automating end-to-end GLR report summarisation via AWS Nova Pro cross-document analysis, removing manual review effort.
• Claim Metrics & QA Audit Chatbots — Two production chatbots on a dual-mode backend: fault-tolerant SQL execution with seamless Pandas fallback for uninterrupted query availability.
• Trust & Visualisation — Validation toggle giving users real-time visibility into AI accuracy, plus Plotly charts generated from natural-language prompts.
• AI Speech-to-Form — Voice-driven data entry for field adjusters (SpeechRecognition + AWS Nova Pro), converting unstructured audio into structured JSON with custom U.S.-state normalisation.

• Built interactive dashboards that cut manual reporting workload by 60% (10+ hrs/week).
• Churn analytics contributing to a 12% improvement in customer retention.
• Deployed a car price prediction model from my MSc dissertation into the production pipeline.

🧠 SQL Generator using AI Agents — LangChain OpenAI API Agentic system turning natural language into executable SQL, with multi-agent pipelines handling autonomous task delegation and inter-agent collaboration.

📚 AI Knowledge Worker (RAG) — FAISS LangChain pdfplumber RAG pipeline over unstructured personal data with FAISS vector similarity search, enabling conversational access across an entire professional history.

💬 Agentic AI Resume Chatbot — Gradio Async I/O Pydantic Persona-driven chatbot that parses PDF resumes and answers questions in real time. Async I/O for concurrency, Pydantic for strict response validation.

🛍️ Fine-Tuned Customer Service Assistant — Transformers PEFT Transformer fine-tuned on retail data for higher-precision customer queries, with streamlined low-latency inference.

🎙️ Automated Meeting Minutes — Whisper Transformers PyTorch End-to-end audio pipeline transcribing meetings and surfacing action items and key decisions from raw audio.

🏥 Medical Insurance Cost Prediction — Scikit-learn Pandas Linear Regression Predictive model estimating insurance charges from age, BMI, smoking status and geographic region, with feature analysis identifying the strongest cost drivers.

⚙️ Predictive Maintenance — Remaining Useful Life — Scikit-learn Time-Series Estimates the Remaining Useful Life of machinery components from sensor telemetry, enabling proactive maintenance scheduling and reduced unplanned downtime.

🎬 Netflix Data Analysis — Pandas Matplotlib Seaborn Exploratory analysis and visualisation of catalogue trends across genres, regions, release years and content ratings.

🚗 Car Price Prediction — Scikit-learn Feature Engineering MSc dissertation research on used-vehicle valuation, later deployed into the production pipeline at Jamjar.co

1. BE Mechanical Engineering — Sri Venkateswara College of Engineering (2019 – 2023)
2. MSc Data Science — University of Essex, UK (2023 – 2025)

1. Master Data Science Program (IIT Madras)

2. The Complete Agentic AI Engineering Course

3. LLM Engineering: Master AI & Agents

4. Generative AI for Software Development & System Design

5. Team Software Engineering with AI

6. The Complete Agentic AI Engineering Course (2025)

7. LLM Engineering: Master AI, Large Language Models & Agents

8. Introduction to Generative AI for Software Development

9. SQL for Data Analysis: Beginner MySQL Business Intelligence

10. The Joy of computing using Python

11. Introduction to Industry 4.0 and Industrial Internet of Things

12. Team Software Engineering with AI

Get in touch · Email · WhatsApp