Skip to content
Nadeem Akhtar

Work

Three systems I designed and built at Synaptics. Each one has the problem, what I did, the result, and the stack.

Synaptics

AP Invoice Automation platform

AP Invoice Automation: enterprise architectureAP invoices input → Oracle Integration Cloud ingestion → Azure Content Understanding extraction → OCI Functions transformation → Oracle staging + AP interface tablesAP invoicesinputOracle IntegrationCloudingestionAzure ContentUnderstandingextractionOCI FunctionstransformationOracle staging +AP interface tables
Architecture
Problem
Automate Accounts Payable invoice processing, from ingestion to Oracle AP interface tables.
Approach
  • Built the original production POC with a live KPI dashboard.
  • Designed the enterprise stack after the POC: Oracle Integration Cloud for ingestion, Azure Content Understanding for extraction, and OCI Functions to transform data into Oracle staging and AP interface tables.
Result
  • Live for the Korea division; rollout underway across 8 global regions.
  • Presented at Synaptics AI Day and recognized as Best AI Work.
Stack
  • Python
  • FastAPI
  • Next.js/React
  • Oracle DB
  • Redis
  • Graph API
  • Oracle Integration Cloud
  • Azure Content Understanding
  • OCI Functions

Synaptics

Scheduled CAD data extraction pipeline

Scheduled CAD data extraction pipelineCAD data input → Alteryx Server scheduled → Embedded Python → Azure Document Intelligence custom model → JSON / CSV outputs → Jira costing + price negotiationsCAD datainputAlteryx ServerscheduledEmbedded PythonAzure DocumentIntelligencecustom modelJSON / CSVoutputsJira costing +price negotiations
Architecture
Problem
Pulling data out of CAD files was a week-long manual process.
Approach
  • Built a scheduled pipeline on Alteryx Server with embedded Python.
  • Trained a custom Azure Document Intelligence model to extract the fields.
Result
  • Replaced a week-long manual process with a few-day automated flow.
  • JSON/CSV outputs feed Jira costing and price negotiations.
Stack
  • Alteryx Server
  • Python
  • Azure Document Intelligence (custom model)

Synaptics

Read-only procurement chatbot

Read-only procurement chatbot in Oracle EBSQuestion in Oracle EBS plain language → Oracle federated Microsoft auth + RBAC → FastAPI service Python, Redis → Read-only SQL agent → Oracle EBS data read-onlyQuestion inOracle EBSplain languageOracle federatedMicrosoft auth+ RBACFastAPI servicePython, RedisRead-onlySQL agentOracle EBS dataread-only
Architecture
Problem
Let staff ask about purchase orders in plain language inside Oracle EBS, with read-only access.
Approach
  • Embedded a chatbot in Oracle EBS that turns plain-language questions into read-only SQL.
  • Access is controlled with RBAC and Oracle federated Microsoft auth.
Result
  • Answers plain-language questions on PO status, open POs by department, and pending approvals.
Stack
  • Python
  • FastAPI
  • Redis
  • RBAC
  • Oracle federated Microsoft auth
  • Read-only SQL agent

Experience

  1. June 2026 – PresentIndependent contractor

    AI Researcher & Engineer, Synaptics

    Bengaluru, India

    • Led end-to-end design and rollout of the AP Invoice Automation platform.
    • Built a read-only procurement chatbot embedded in Oracle EBS.

  2. June 2025 – June 2026Internship

    AI/ML Intern, Synaptics

    • Built the AP Invoice Automation POC.
    • Built the CAD extraction pipeline.
    • Integrated Oracle Digital Assistant into Oracle EPM.

  3. July 2022 – Nov 2022Internship

    Python Developer Intern, Qedge Tech

    • Built backend applications with Python, Flask, and Django: APIs, business logic, and database integrations.