Builder mindset
I take ideas from rough requirements to documented, tested products with clear API, data-flow, and deployment decisions.
AI & Security Engineering / Cisco / Bengaluru
Manohar Eldhandi
Software Engineer
At Cisco, I build Python agentic-AI workflows for code-security reviews. The security-compliance tooling is used by 50+ engineers across 4 product teams and has reduced manual review effort by about 70%.
CodeChef 4-star
Codeforces Master - 2141
Amazon ML Summer School top 1% of 50,000+
50+ engineers across 4 Cisco product teams
11-stage pipeline / 10x fewer model calls
736 pytest tests / 97% UI coverage
4 ATS connectors / 120 curated boards
LER_DSA used by 500+ juniors
About
A developer who learns fast, builds reliable systems, and owns the outcome.
I am a Computer Science graduate and Cisco Software Engineer focused on AI & Security Engineering. My work spans Java and Python backend systems, deterministic-first AI workflows, MCP integrations, GraphQL and API design, audit evidence, and browser-based release automation.
What makes me useful on a team is the mix of fundamentals, ownership, and calm debugging. I can step into an unclear problem, narrow risk, make the system easier to reason about, and ship something another engineer can review, trust, and extend.
Backend depth
I think in APIs, schemas, access control, state transitions, auditability, failure modes, and maintainable handoff.
Practical AI/ML
I build LLM and ML workflows around retrieved evidence, structured outputs, measurable evaluation, and decisions people can inspect.
Reliability focus
I care about traceability, automated tests, CI/CD, review evidence, and release validation because they make systems easier to trust.
Experience
Where I learned to build, explain, and improve.
Oct 2025 - Present
Software Engineer
Building Python agentic-AI systems for security-compliance workflows at Cisco. The work spans structured code review, model orchestration, MCP integrations, grounded retrieval, and automated release validation.
- Designed and built 3 Python agentic-AI workflows that generate structured JSON for code-security reviews with LLM APIs and prompt engineering, cutting manual review effort by about 70% across 4 product teams.
- Used LangChain to wrap 2 LLM providers, Codex and GitHub Copilot, behind one chat interface and LangGraph to orchestrate agents over them without provider-specific workflow changes.
- Integrated multi-provider MCP tool servers for security data and Jira into LangGraph agents, then designed reusable Agent Skills for 50+ engineers across 4 teams.
- Assisted with a RAG grounded-retrieval layer and code graph for 20K+ source files; 50+ parallel workers feed an 11-stage pipeline that cut model calls 10x.
- Built Playwright UI tests at 97% coverage and executed 736 pytest tests across 10+ CI/CD releases, reducing manual QA by 60%.
Feb 2024 - Jun 2024
DSA Mentor
Mentored juniors in DSA by helping them understand patterns instead of memorizing solutions. The work improved how I explain logic, break down problems, and guide someone from confusion to a clean implementation.
- Guided 150+ students through debugging, optimization, and interview-style problem solving.
- Helped students improve problem-solving by 40% on average and earned Smart Coder top 2% national recognition.
Dec 2021 - Jul 2025
B.Tech, CSE with AI and ML
Built the foundation I still rely on: DSA, OOP, DBMS, operating systems, computer networks, software engineering, and applied machine learning. College gave me the base; projects and Cisco made it practical.
- Graduated with an 8.5/10 CGPA in Computer Science with AI and ML.
- Used coursework and projects to connect backend engineering with data and ML systems.
Selected builds
Projects that show how I think and build.
01
Full-stack product
OnTheWay - Route-Aware, Event-Driven Pickup Platform
A Spring Boot and React platform for route-aware pre-ordering across customer, merchant, and admin roles. Its ETA-synchronization engine uses a pluggable RouteProvider to time preparation so an order is ready when the customer arrives.
The current build adds 47 REST endpoints and 3 GraphQL operations, JWT/BCrypt authorization, Kafka order events, Elasticsearch catalog search, Stripe/Razorpay adapters, Flyway migrations, Docker, Kubernetes, CI, 85 JUnit tests, and a developer-machine 1,000-user load test that sustained 730 sessions per second at p99 884 ms with zero failures.
How it works
Ready exactly on arrival
Find nearby shops and items by route, distance, and vertical.
Checkout calculates travel time, prep duration, and safety buffer.
Live location updates keep merchant preparation aligned to arrival.
The order reaches READY as the customer reaches the store.
RealtimeWebSockets
SearchGraphQL + ES
EventsKafka
DeliveryDocker + K8s
115 shops / 507 items
16 verticals
47 REST + 3 GraphQL
Kafka + Elasticsearch
85 tests / 730 sessions per sec
GitHub repo
02
Local-first AI product
Carivyo-AI - Local-First Career-Intelligence Platform
A local-first React and FastAPI product for resume-led job discovery and application prep. It supports PDF, DOCX, TeX, Markdown, and text intake, adaptive interviews, evidence-backed matching, and profile-aware job ranking.
The current build uses 2 model runtimes, 4 official hiring-system connectors, 120 audited company boards, a first-party 27-check ATS diagnostic, explicit provenance, redacted logs, prompt-injection defenses, and approval gates that invalidate stale evidence.
4 ATS connectors
120 curated boards
27-check ATS
2 runtimes + approval gate
GitHub repo
03
Django ML application
NeoPulse Risk - Django Cardiac Risk Screening App
A Django machine-learning application for cardiac risk screening experiments. It collects 12 structured indicators, runs a reusable scikit-learn prediction service, returns inline lower/elevated-risk output with probability, and publishes model metrics on the About page.
It keeps reports in the visitor's signed browser session instead of a public history, supports downloadable text reports, uses grouped field guidance for cleaner input, includes tests and CI, and is ready for Render deployment.
12 input features
Probability + metrics
Session-only reports
Render-ready
GitHub repo
Skills
A stack built around systems that ship.
AI
& Security
& Security
Java
Python
Spring
FastAPI
GraphQL
Kafka
LLM APIs
Codex
MCP
RAG
sklearn
LangGraph
Git
Docker
K8s
CI/CD
Playwright
GH Actions
Proof
Proof that the work is backed by results.
CodeChef 4-star and Codeforces Master
Reached 1893 on CodeChef and 2141 on Codeforces, with global ranks 32 in Starters 239, 140 in Educational Codeforces Round 191, and 295 in Codeforces Round 1100.
Amazon ML Summer School 2024
Selected in the top 1% from 50,000+ applicants, strengthening my understanding of ML modeling, evaluation, and practical AI systems.
Cisco Webex Playtime Top 35
Ranked Top 35 among 2,000+ teams by building an AI-based security compliance automation tool in a fast, team-driven setting.
11-stage security review pipeline
Assisted with a grounded-retrieval and code-graph pipeline for 20K+ source files; 50+ parallel workers reduced model calls by 10x while preserving provenance.
150+ students guided in DSA
Mentored students in DSA, helped improve problem-solving by 40% on average, and earned Smart Coder top 2% national recognition.
LER_DSA adopted by 500+ juniors
Published an MIT-licensed 30-day Java interview roadmap with 20 topic modules, reusable templates, progress tracking, and clean documentation used across college prep groups.
Contact
Looking for a software engineer who can connect backend depth with AI and security engineering?
I am open to software engineering conversations around backend systems, AI and security products, developer tooling, test infrastructure, and teams where ownership, evidence, and release reliability matter.