Aisha Siddiqua
I build AI agents that take manual work off your team.
New to Contra
Aisha is ready for their next project!
My end-to-end approach to shipping AI agents: voice or chat interface, an agent layer with tool-calling and guardrails, deterministic logic for anything that must be auditable, and persistent memory in PostgreSQL + pgvector.
Built on Python 3.11, FastAPI, and OpenAI Agents SDK. Anthropic MCP certified.
I also lead applied AI instruction for a government-recognized national program, teaching this same stack to working engineers.
0
8
An agent that monitors open cases against deadlines, moves them through a state machine, and escalates automatically when a threshold is crossed. Removes the manual chase-up work a support team does by hand every day.
Stack: Python, FastAPI, PostgreSQL.
1
16
A retrieval layer that lets an agent recall a user's history across separate conversations, using pgvector inside PostgreSQL — no separate vector DB to run or pay for.
Stack: PostgreSQL, pgvector, text-embedding-3-small, Python.
Why it matters: most chatbots forget you the moment you close the tab. This makes an agent feel continuous.
0
18
uilt a voice-first AI agent that takes customer complaints by speech or text in two languages, classifies them, routes them to the right team, and tracks resolution status across sessions.
Stack: Python, FastAPI, OpenAI Agents SDK, Whisper STT, PostgreSQL + pgvector.
Key engineering: deterministic routing logic instead of LLM calls, so every decision is auditable. Cross-session memory retrieval via vector search — the agent remembers a user's prior issues. Single-mutator write discipline for a clean audit trail
0
22