Kalyanasundharam Kathiresan
Software Engineer 2 · Apptium Northern Virginia · Java · AWS · AI tooling

Kalyanasundharam Kathiresan
Backend engineer.

Software engineer working across backend systems and AI tooling. I modernize Java microservices, make slow APIs fast, and build the AI agents engineering teams use every day. 6.5+ years across commerce, media, and enterprise SaaS.

Impact log

Apptium2025 – now
−67%API latency on core commerce services, through caching, parallel calls, and removing a middleman service
20 minTypical issue triage, down from 2–3 hours, with AI log analysis I built
5 teams~40 engineers adopted my AI story-grooming agent
Brightspot2023 – 2024
10M+Articles and media moved in a four-month CMS migration I led
3Enterprise client platforms upgraded without losing their customizations
DemandFarm2018 – 2021
4×QA, Sales, and Customer Success efficiency from a demo builder tool
−35%Page load times with reusable Lightning Web Components
30+Users on a program management app I built end to end

Selected work

Problems I owned end to end, what changed because of them, and how each one works.

Systems I rebuilt

−67%p90 latency on tax calculation
Apptium · commerce platform

Tax calculation, rebuilt in-service

Tax requests used to hop through another team's sunsetting middleman service. I cleared its bottlenecks and moved the logic in-house, so the service now calls the third-party tax API directly.

JavaQuarkusREST
Before and after, request path
−33 to 50%p90 latency on core APIs
Apptium · commerce platform

Quarkus migration and faster APIs

Migrating a core commerce service from legacy Spring Boot to reactive Quarkus on AWS, keeping both at feature parity. Upstream calls now run concurrently, and repeat lookups come from Redis.

QuarkusMutinyRedisAWS
Sequential vs. parallel calls (illustrative timings)
1 → None edit fans out to every dependent record
Brightspot · data model

Cascading updates engine

An edit to one record queues each dependent record in a graph-like data model. Worker threads drain the queue in parallel, and shared dependents are queued only once.

JavaMulti-threadingQueues
Queue and worker threads
10M+articles and media migrated
Brightspot · news agency

10M+ article CMS migration

Led a four-month, three-phase migration for a major news agency: articles, media, then press releases, from 10+ PostgreSQL tables and multiple GCS buckets.

JavaJOOQPostgreSQL
Three migration phases
4×QA, Sales, and Customer Success efficiency
DemandFarm · internal tool

Demo builder

Demo setup that three teams did by hand, task by task, became a single click.

SalesforceApexLWC
Manual setup vs. demo builder
−35%page load times
DemandFarm · front end

Reusable component library

One shared set of Lightning Web Components built every page and standardized the design across the product.

LWCSLDSAura
Shared components assembling pages

AI tools I built

20 mintypical triage, down from 2–3 hours
Apptium · AI agents

Copilot agents for engineering teams

A story-grooming agent that writes Jira stories from chat, used by 5 teams (~40 engineers), plus agent skills that analyze Kibana logs to root-cause bugs.

CopilotKibanaJira
~160repos indexed across 3 GitHub orgs
Apptium · internal tool

Code intelligence platform

Search, definitions, references, and call graphs through an API, CLI, browser UI, and an MCP server. Used by executives and leads to trace cross-team dependencies.

Pythontree-sitterMCP
Adoptedbuilt into production by another team
Apptium · prototype

Conversational ordering agent

An agent that lets retail and telesales agents place orders by chat. Another team took the concept and built it into a dedicated ordering channel.

Google ADKPythonLLMs

Experience

Six and a half years across commerce, media, and enterprise SaaS.

Mar 2025 – now
Reston, VA

Apptium Technologies

Software Engineer 2 · embedded with Comcast's Xfinity Mobile team
  • Migrating a core Xfinity Mobile commerce service from Spring Boot to reactive Quarkus on AWS.
  • Cut p90 latency by 33–67% on core APIs with parallel upstream calls, Redis caching, and refactoring.
  • Authored 6 design documents, defined API contracts for dependent teams, and wrote ~30 stories across 5 epics.
  • Built AI tooling adopted beyond my team: a story-grooming agent, log-analysis skills, and a code intelligence platform.
  • Backup scrum facilitator for the team.
Feb 2023 – Nov 2024
Reston, VA

Brightspot (Perfect Sense)

Software Engineer 2 · Software Engineer 1
  • Designed a queue-based, multi-threaded engine for cascading updates in a graph-like data model.
  • Led a 10M+ article migration for a major news agency.
  • Built GraphQL endpoints feeding a partner's recommendation model, and ran CMS upgrades for 3 client implementations.
May – Aug 2022
Manassas, VA

Berkley Technology Services

Software Developer Intern
  • Built Dynatrace dashboards that helped cut downtime by 15%, and integrated alerts with Teams and ServiceNow through Flask APIs.
Jul 2018 – Jun 2021
Pune, India

DemandFarm

Software Engineer
  • Built a full-stack program management app adopted by 30+ users at a key customer.
  • Cut load times 35% with reusable Lightning Web Components, and built a demo tool that made QA, Sales, and Customer Success 4x faster.

Skills

Languages

JavaPythonSQLJavaScriptTypeScript

Backend

QuarkusSpring BootSmallRye MutinySpring SecurityHibernateRESTGraphQLOAuth2

Data & messaging

RedisDocumentDBOraclePostgreSQLMongoDBKafkaKinesis

Cloud & delivery

AWSDockerKubernetesConcourseGitHub ActionsMaven

Observability

KibanaDatadogDynatrace

AI tooling

Copilot agentsMCPGoogle ADKClaude APItree-sitter

Earlier projects

Education

University of Virginia

Master of Computer Science · 2021–2022

MGM's Jawaharlal Nehru Engineering College

B.E., Computer Science and Engineering · 2014–2018

Let's build something fast.

Open to conversations about senior backend and platform engineering roles.