hey, i'm aadhil imam

Jul 2026 – Present · Dubai, UAE

• Built Gabi Studio, Sparkit's new AI design assistant, from scratch and took it to production on web, iOS, and Android, replacing the previous design agent as the main creative surface.
• Designed its multi-agent architecture: an orchestrator that works with the user and calls on research and analyst specialists, with every step graded by a separate judge model and repaired once if it fails.
• Cut the cost of a simple request by 70% while keeping performance consistent.
• Took users from a text prompt, catalogue item, or photo to a priced, production-ready tech pack in one chat, including multi-piece collections rendered in the background with automatic logo checks.
• Built an agent evaluation platform to compare models, run A/B tests, and evaluate the full system, and added evaluation loops inside the agent.
• Added memory layers so the agent recalls past designs, prices, and preferences across conversations, reusing earlier work at no extra cost.
• Built an agent ops platform to monitor, manage, and trace every agent, with per-turn cost and quality tracking, plus social media publishing and a behavior-based notification aggregator.

Feb 2025 – Jun 2026 · Dubai, UAE

• Developed a multi-agent Legal AI platform that automates document analysis, legal research, drafting, and client interactions, with an LLMOps-based evaluation and monitoring framework for production reliability, deployed as scalable FastAPI microservices on containerized infrastructure.
• Designed and deployed a knowledge-driven CRM chatbot using an advanced RAG architecture and LangChain to deliver accurate, context-aware responses from CRM data, leading its end-to-end development and production maintenance.
• Built a GenAI-powered credit portfolio analyzer that automates debt portfolio insights and risk prioritization, reducing manual review time by 45% and improving collection efficiency by 35%.
• Developed a Legal Matter Analyzer integrated with the CRM using LangChain, Azure OpenAI, Django, and PostgreSQL, delivering intelligent classification and recommendations with 92% accuracy and cutting processing time by 40%.
• Architected production-grade GenAI microservices for email automation, sentiment analysis, timesheet summarization, and speech-to-text pipelines, improving operational productivity by 38% and cutting manual workload by 50%.
• Built an AI-powered outbound voice calling agent using Twilio and OpenAI with real-time text-to-speech, enabling human-like automated conversations for customer outreach.
• Engineered a Power BI analytics dashboard with real-time KPIs and a scalable FastAPI analytics microservice for dynamic data visualization across multiple sources.

Dec 2021 – Jan 2025 · Colombo, Sri Lanka

• Developed an LLM-powered microservices generator and SRS document generator using LangChain, Azure OpenAI, and advanced prompt engineering, significantly improving developer productivity and code generation quality.
• Designed and implemented a RAG-based document chatbot enabling interactive Q&A on uploaded documents.
• Built customer segmentation clustering models that reduced marketing expenditure by 30% through precise targeting, and implemented a credit scorecard model with a dynamic credit limit decision engine.
• Established MLOps pipelines on AWS SageMaker for automated model training and deployment, and built AutoML workflows with PyCaret.
• Designed cloud-based ETL pipelines (Azure Data Factory, Databricks, Spark, Airflow, Snowflake) for forecasting, an information security data lake, and modernized KPI reporting.

Nov 2020 – Dec 2021 · Colombo, Sri Lanka

• Performed customer segmentation based on buying power to identify high-potential targets and optimise marketing strategies.
• Developed a rule-based product preference prediction model for non-sachet users, increasing sales by 60%.
• Researched and engineered a deep learning recommendation engine achieving 85% accuracy on 500,000 customer records for new customer targeting.
• Conducted exploratory data analysis (EDA) on customer behavioural data to deliver actionable insights to product teams.