Skip to content
Awsaf Alam
GitHubLinkedIn

Work

We turn messy workflows into AI products people can use. The first story is our current work at Sourcy: the problem, what we did, and where it stands now. Other 2026 R&D, earlier platform work, open source and research follow. Each case study has a small interactive demo, labelled as a simulation, reconstruction or design study.

Current work

  • Supplier conversations that can be inspected and improved

    Problem
    A sourcing request can't become a useful quotation until suppliers answer, and the agent asking them was hard to inspect or fix.
    Our contribution
    Rebuilt the agent around explicit goals, conversation history and human escalation, with end-to-end tracing and regression testing.
    Read the case study: Supplier conversations that can be inspected and improved

Other R&D (2026)

  • Designing memory for agents

    A design for how agents remember: what to keep, what to forget, and how to find it again. Research and design, not shipped.

Platform & product (2023–2025)

  • AI insights for TikTok and Instagram

    A social media insights platform where AI agents work on data that never stops arriving.

  • Smart catalog search

    Customers get sourcing answers from a ready catalog instead of waiting for a search to run.

    • 90% shorter customer wait times for sourcing results, through intelligent search and proactive data scraping

  • Scaling and reliability

    A backend that kept up with fast growth and went down less often.

    • 30% month-over-month growth in transaction volume that the infrastructure was scaled to support
    • 25% less platform downtime after introducing monitoring
    • 40% faster database queries through normalization, indexing and caching

Early-stage startup (2020–2023)

Projects

Research

© 2026 Awsaf Alam