Skip to content
Deepanshu Kr Chaurasia

Deepanshu Kr Chaurasia · Software engineerI design, build and run secure backend systems.

Python, Django and .NET on Azure, with sign-in, permissions and audit designed in. Nearly three years building a healthcare platform.

ClearInsight AI website preview introducing its dental referral platform.
SC-Oracare website preview with an example dental practice dashboard.
P-Buddy website preview featuring its certification practice dashboard.

Engineering across products and platforms

— Healthcare, AI, automation, and learning.

sc-oracare.onrender.com
The SC-Oracare dashboard on demo data: referrals needing attention, today’s schedule and the completion overview
  • Referral network: referrals, treatment teams and live treatment rooms across practices.

    Load-tested at 100+ concurrent WebSocket connections.

    Read the case study about ClearInsight AI
  • Practice management: scheduling, charting, revenue cycle and a patient portal, around the referral core.

    Runs today as a live demo on synthetic data.

    Read the case study about SC-Oracare
  • Practice platform: question banks as content, server-owned exams and progress backed by evidence.

    Live at pbuddy.study with SC-300; DP-300 and resume-based interview preparation are built and open next.

    Read the case study about P-Buddy
Open live demo (opens in new tab)

Hosted on a slower server; the first load can take up to a minute.

Oracare AI

Clinical AI and retrieval service

Problem
Clinicians need answers from one patient’s records without an AI service ever seeing identifiers or other patients.
What I built
A FastAPI service for dictation, document capture and retrieval with citations, isolated per patient by row level security.
Proof
Retrieval is built and tested; generation is the next phase.
  • FastAPI
  • PostgreSQL
  • pgvector
  • Row level security
  • Microsoft Entra ID
  • Azure Container Apps
Identity providerAzure Container AppsSpeech and document providersRetrieval indexSC-OracareServer side onlyMicrosoft Entra IDTokens, signing keysAPI containerFastAPI, validates every callModel runtimeHolds the modelsSpeech to textAzure AI Speech or DeepgramAzure Document IntelligenceDocument capturePostgreSQL with pgvectorForced row level security

Simplified architecture of Oracare AI as it runs today, called only by SC-Oracare. The full walkthrough is in the case study.

CI-Assistant

Windows connector and plugin platform

Problem
Practice software runs on Windows PCs inside clinics, out of reach of a cloud API.
What I built
A pull-only Windows connector that signs in with JWTs and runs signed, hash-verified packages in a sandbox.
Proof
Architecture designed and prototyped; integrations explored, not shipped.
  • Python
  • C#
  • .NET
  • JWT
  • Azure Queue Storage
  • Azure Blob Storage
  1. Published
  2. Queued
  3. Signed link
  4. Downloaded
  5. Hash verified
  6. Sandboxed

Publish a package to step through each stage.

Earlier projects

Phishing Domain Detection

Machine learning pipeline on Azure

Problem
Phishing links imitate real domains closely enough to pass a quick look.
What I built
A classifier with features engineered from URL structure, domain patterns and character distribution.
Proof
Trained with scikit-learn and pandas, and deployed on Azure.
  • scikit-learn
  • pandas
  • Azure

Image to Text Converter

OCR web application

Problem
Text and context inside images cannot be searched, edited or reused.
What I built
A Flask application that extracts text with Tesseract OCR and describes scenes with TensorFlow.
Proof
Each result exports as PDF, TXT or DOCX.
  • Flask
  • Tesseract OCR
  • TensorFlow

How I build

Each principle comes with a small working demo.

Know who is asking

Pick a role and watch the sign-in path: the token carries the role, the policy decides, and the audit log records it.

Sign in as
  1. Browser to Entra External ID: Sign in (OIDC)
  2. Entra External ID to Browser: Authorization code
  3. Browser to Django API: Callback with code
  4. Django API to Entra External ID: Exchange code
  5. Entra External ID to Django API: Tokens (JWT)
  6. Django API to RBAC policy: Check role
  7. RBAC policy to Django API: Permitted
  8. Django API to Audit log: Write audit entry
{
  "iss": "https://example.ciamlogin.com",
  "aud": "example-api",
  "sub": "example-user",
  "roles": ["practice-administrator"],
  "exp": 1767225600
}

Practice administration: Permitted

Simplified ClearInsight AI sign-in path. Example policy, not the production rule set.

Isolate data where it lives

Search one patient’s records. Row level security in PostgreSQL decides what comes back, so even a query with no patient filter stays inside one chart.

Simulated with synthetic records.

Patient
Search for
-- Patient P-1042 set for this transaction only,
-- from the access grant
SELECT text, source FROM records
WHERE topic = 'allergies';  -- no patient clause

Make it observable

One request through ClearInsight AI, span by span. Logs, metrics and alerts went to Application Insights.

Example trace: span order only, no timings.

Select a span to see what happens there.

Ship through a pipeline

Every change takes the same path from commit to a deployed environment, through GitHub Actions and Azure DevOps pipelines.

Run through GitHub Actions and Azure DevOps pipelines.

  1. Commit
  2. Tests
  3. Container build
  4. Deploy

Push a change to step through each stage.

Experience

Education, work and projects over time
ActivityPeriodTimeline
Bachelor of Computer ApplicationsHimachal Pradesh UniversityJul 2016 to Jul 2019
Competitive examination preparationJul 2019 to Jul 2021
Master of Computer ApplicationsChandigarh UniversityJul 2021 to Jul 2023
Data Science InternAllHeart WebJan 2023 to Jul 2023
Software EngineerShidulAug 2023 to Jun 2026
SC-OracareCloud dental practice management platformJun 2026 to present
P-BuddyPersonal practice and performance companionSep 2026 to present

Diamonds mark certifications earned in 2022.

Remote, for a US healthcare technology client

Contract engagement. Software engineer and architect in the client’s four-person team, across architecture, backend, identity and security, cloud infrastructure and release automation. Built two complete systems for a dental healthcare startup, which ceased operations before commercial launch; both systems run today as live demonstrations.

ClearInsight AI, dental referral and workflow platform

  • Designed and built the identity and access layer: Azure AD B2C and Microsoft Entra External ID over OAuth 2.0 and OpenID Connect, JWT session security, and role based access control enforcing least privilege from referring dentist to specialist to practice administrator.
  • Architected the platform in Python, Django, PostgreSQL and JavaScript, with Django REST Framework APIs for referral routing, patient records, document exchange and practice administration.
  • Built to HIPAA requirements: TLS in transit, encryption at rest, an authorization check on every patient data operation, and an audit trail of who accessed or changed each record.
  • Delivered live referral status and notifications over Django Channels, Redis and WebSockets, verified at 100+ concurrent connections with Apache JMeter and Azure Load Testing.
  • Containerized with Docker, and covered authentication, RBAC enforcement and referral state transitions with pytest unit and integration tests.
  • Instrumented structured logging, Azure Application Insights metrics and alerting for request failures, authentication errors and WebSocket connection health.
  • Provisioned and operated nine Azure services, including App Service, Blob and Queue Storage, Front Door, Key Vault and Azure Database for PostgreSQL, with GitHub Actions and Azure DevOps pipelines from commit to deployed environment.
Read the case study about ClearInsight AI, dental referral and workflow platform

CI-Assistant, Windows connector and plugin platform

  • Designed and built connector-based desktop and backend solutions in Python, C# and .NET for plugin execution and workflow automation in dental practice software.
  • Secured API and WebSocket communication with JWT, signed URLs and hash-based verification, using Azure Queue Storage and Blob Storage for asynchronous package delivery.
  • Designed and prototyped a plugin-based connector architecture, exploring integration with several dental practice management systems through API research, desktop automation experiments, secure messaging and isolated plugin execution.
Read the case study about CI-Assistant, Windows connector and plugin platform

What I work with

Identity and access

Enterprise sign-in and least-privilege access in every product, from the referral platform to the AI service.

Tools: select one to see where it was used

Used in

Education and certifications

  • MCA, Chandigarh University

    · Cloud computing and DevOps · GPA 8.11 / 10

  • BCA, Himachal Pradesh University

Also completed: Claude Code in Action (opens in new tab) (Anthropic, 2026), Fundamentals of Red Hat Enterprise Linux (opens in new tab) (Red Hat, 2022).

About

Deepanshu Kr Chaurasia

I’m a backend engineer in Chandigarh, India. I like owning a system end to end: how people sign in, what each role may see, how data stays isolated, and how it all gets tested, shipped and watched in production.

For nearly three years I was software engineer and architect on a US dental healthcare product, working with the client’s product owner, an RPA and plugin solutions engineer, and a marketing and customer relations lead. When the client closed, I kept building: the platform is now SC-Oracare, with Oracare AI alongside it.

I’m looking for a backend, identity or AI platform role, remote or with relocation.

How I work

  • Design the permission model first.
  • Prefer boring, observable infrastructure.
  • Write the decision record before the code.
  • Ship small, behind flags.

Get in touch

Open to backend, identity and AI engineering roles, remote or with relocation.