Skip to content
View Aparnap2's full-sized avatar
🖥️
working on portfolio projects
🖥️
working on portfolio projects

Block or report Aparnap2

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Aparnap2/README.md

Hi, I'm Aparna Pradhan 👋

Code‑First Automation Architect | Full‑Stack & AI Agents

"Outcomes over demos. Governance over hype."

I modernize brittle, legacy/no‑code automations into reliable, scalable, cost‑efficient systems. I don't just write scripts; I build production-ready architectures with binary acceptance tests, P95 latency targets, and live ROI dashboards.


🛠️ The "Senior" Stack

I bridge the gap between AI research and Enterprise reliability.

Category Technologies
Language Python TypeScript
AI Orchestration LangGraph LiteLLM RAG (Graph/Vector)
Backend FastAPI NestJS Pydantic
Data & Vector PostgreSQL Redis Neo4j
Infrastructure Docker Langfuse GitHub Actions

🚀 Flagship Engineering Capabilities

I build systems that hit specific Service Level Objectives (SLOs).

1. Finance Inbox & Procurement (AP/AR)

Automated reconciliation with 100% duplicate detection.

  • Scope: Docling OCR + Pydantic validators + Anomaly digests.
  • Metric: ≥98% field accuracy on 200‑doc test sets.
  • Stack: Python, FastAPI, GraphRAG.

2. Generative Support Workforce

Email/WhatsApp resolution with strict governance.

  • Scope: Citations required, QA gating, Sentiment routing.
  • Metric: P95 response < 2 minutes; Breach alerts < 30s.
  • Stack: LangGraph, Redis Queues, LiteLLM.

3. Booking & Lead Ops

Idempotent calendar operations.

  • Scope: Intake → Qualifier → Slot Picker → Reminders.
  • Metric: +20% show‑rate vs baseline; < 60s write latency.

🧠 Engineering Principles

  • Predictability: Typed data flows end‑to‑end (TypeScript/Zod/Pydantic). No "stringly typed" code.
  • Security: Least privilege, audit logs, and PII redaction by default.
  • Observability: If it isn't logged in Langfuse/Phoenix, it didn't happen.
  • Ownership: Code‑first, no lock‑in. I build systems your team can extend.

🤝 Engagement Models

I work best with ops‑minded founders who value clear scope and sustainable systems over throwaway prototypes.

  1. 10‑Day Modernization Audit: Latency/cost baselines, SLAs, and a fixed pilot SOW.
  2. Pilot Build (10–14 days): Pass/Fail delivery based on acceptance criteria.
  3. Ongoing Ops: Monthly SLOs and change-managed improvements.

📬 Connect

Pinned Loading

  1. invoicify invoicify Public

    Vertical AI Agent for Finance Operations - Automated invoice processing with Analyst-Critic pattern, Trust Battery system, and Slack "Intern's Desk" interface.

    TypeScript

  2. personal_assist_app personal_assist_app Public

    AI assistant that produces brand-consistent social posts and organized Notion artifacts, with decisive consultation and human approvals.

    Python 1

  3. smart_commerce_with_vercel-ai-sdk smart_commerce_with_vercel-ai-sdk Public

    A production-ready AI-powered e-commerce support chatbot using Vercel AI SDK, Google Gemini, Prisma, Neon DB, and Zod validation.

    TypeScript 1

  4. devops_agent devops_agent Public

    Autonomous DevOps/SRE Agent with LangGraph orchestration for proactive incident detection, diagnosis, and remediation.

    Python

  5. ExecOps ExecOps Public

    Active agent automation for SaaS founders. Four vertical agents handle domain-specific workflows with human-in-the-loop approval.

    Python

  6. personal-research-agent personal-research-agent Public

    a powerful, AI-driven research assistant that transforms complex research queries into comprehensive, data-driven reports

    Python