KK
Available for opportunitiesAhmedabad, IndiaSoftware Engineer · AI · Backend

KarmaKansara.

Karma Kansara
Portrait
Karma · 2026
Location
IND · GMT+5:30

Software engineer building AI-native products — from agentic systems and event-driven microservices to APIs that quietly do the hard work.

View selected workGet in touch
GitHub ↗LinkedIn ↗
Python✦FastAPI✦Next.js✦TypeScript✦RabbitMQ✦Docker✦vLLM✦Multi-agent LLMs✦Supabase✦ClickHouse✦Pydantic✦XGBoost✦Hadoop✦MITRE ATT&CK✦.NET✦C#✦Python✦FastAPI✦Next.js✦TypeScript✦RabbitMQ✦Docker✦vLLM✦Multi-agent LLMs✦Supabase✦ClickHouse✦Pydantic✦XGBoost✦Hadoop✦MITRE ATT&CK✦.NET✦C#✦
01About

I build the quiet infrastructure
behind AI-native products.

Currently
Software Developer at Ekesa Tech
Focus areas
  • Agentic AI & multi-agent orchestration
  • Event-driven microservices
  • APIs, pipelines, distributed data
  • Full-stack (FastAPI · Next.js)
Based in
Ahmedabad, India · Open to remote
02Experience

Where I've done the work.

Roles I've held and the systems I owned there.

01Experience

Ekesa Tech

Software Developer
June 2025 — Present
Tech
PythonFastAPIRabbitMQDockerNext.jsSupabase
What I did · 4 bullets
  • Engineered an AI-powered resume parsing engine using Python that has processed 115,000+ CVs, eliminating third-party dependency and replacing a fully manual hiring workflow end-to-end.
  • Migrated a monolithic architecture to event-driven microservices using RabbitMQ, building secure system integrations that offload bulk uploads (500+ files) to background queues for reliable, uninterrupted service at scale.
  • Designed and built RESTful APIs in FastAPI with Docker-based deployment, delivering data-exchange pipelines between the ingestion service, parsing engine, and Next.js frontend backed by Supabase auth.
  • Refactored a resource-intensive legacy search query, significantly reducing response latency; collaborated cross-functionally with product and design stakeholders across Agile sprints, scope estimation, and code reviews.
02Experience

Biztic Technologies

Software Development Intern
Jan 2025 — May 2025
Tech
C#SQLXGBoostRandom ForestPython
What I did · 3 bullets
  • Built C# and SQL parsers to extract and normalize metadata across 80,000+ hours of OTT content and 81 live channels, achieving ~95% accuracy with unit tests validating pipeline correctness.
  • Designed and owned end-to-end data integration pipelines — ingestion, cleaning, normalization, transformation — enabling downstream analytics on a previously unstructured content catalogue.
  • Developed a pricing prediction model using XGBoost and Random Forest achieving 85% accuracy, enabling data-driven content acquisition decisions across the business.
03Selected Projects

Things I've built.

Real, shipped, and built to be maintained by humans other than me.

01Project

Aegis

AI-Powered SOC Platform
2025
Tech
PythonvLLM/ROCmStreamlitPydantic
What I did · 4 bullets
  • Built a multi-agent agentic AI platform for SOC triage using 5 specialized agents (Parser, Correlator, Triage, Planner, Reporter) that reason through security alerts and produce auditable verdicts — splitting a monolithic prompt into narrow agents for legible, traceable reasoning.
  • Designed a human-gated workflow where agents recommend and humans approve — no auto-execution — ensuring secure, privacy-compliant operation with zero data leaving the environment on threat-intel lookups.
  • Implemented 15 detection rules with automatic mapping to MITRE ATT&CK tactics, generating deployable Sigma, YARA, Splunk SPL, and Elastic queries for downstream SIEM integration.
  • Benchmarked concurrent throughput on AMD GPU (vLLM/ROCm) vs hosted API — measuring alerts/sec, p50/p95 latency, and cost-per-1k-alerts — to validate scalability of the agentic pipeline under load.
02Project

Audit Sphere

Governance, Risk & Compliance Tool
2024
Tech
C#.NETMySQL
What I did · 3 bullets
  • Designed and built a modular desktop GRC application for ISO 27001 & NIST compliance, replacing manual spreadsheet-based audits with a structured, automated workflow supporting control mapping, auditor role assignment, and full audit lifecycle execution.
  • Implemented role-based access control and automated report generation, ensuring only authorised personnel can view or act on sensitive compliance data.
  • Delivered a real-time compliance dashboard with visual tracking metrics; covered core modules with unit tests to ensure correctness across framework mappings.
03Project

Sentiment Analysis on Twitter

Demonetization Study
2023
Tech
HadoopApache PigPythonNLP
What I did · 2 bullets
  • Processed a 20GB Twitter dataset on a Hadoop/Cloudera cluster, writing Apache Pig scripts for parallel distributed transformation and ingestion of large-scale unstructured social data.
  • Applied NLP-based sentiment classification to categorise public opinion on India's 2016 demonetization policy; generated trend visualizations to surface key discourse patterns across time.
04Toolkit

The stack, honestly used.

01
6 items

Languages

PythonC / C++C#JavaScriptTypeScriptSQL
02
7 items

Backend & Frameworks

FastAPI.NETNode.jsNext.jsRESTful APIsMicroservicesRabbitMQ
03
6 items

AI / Agentic

Multi-agent SystemsLLM OrchestrationvLLMOpenAI APIPydanticRAG Pipelines
04
3 items

Databases

MySQLClickHouseSupabase
05
7 items

ML & Data

XGBoostRandom ForestNLPPandasNumPyHadoopApache Pig
06
5 items

DevOps & Practices

DockerGitpytestAgile / ScrumCode Reviews
05Education
DegreeJune 2021 — May 2025

Rashtriya Raksha University

B.Tech in Computer Science & Engineering
Gandhinagar, India
Beyond the classroom

Hackathons, agentic AI research, and shipping side-projects that teach me the parts a syllabus can't — production trade-offs, on-call reasoning, and how software breaks in the wild.

06Contact
Let's build something

Have a hard
problem?

I'm open to full-time roles, contract engagements, and collaborations on AI systems, backend infrastructure, and full-stack products. The best way to reach me is by email.

karmakansara@gmail.comDownload Résumé
Email
karmakansara@gmail.com
Phone
+91 97277 78766
LinkedIn
/karma-kansara
GitHub
/arthados
Location
Ahmedabad, India