01 · IMPACT
Same workflows.
Minutes, not days.
Agent systems shipped for paying enterprise customers in finance, healthcare, supply chain, and security.
01
4 days → 30 min
Enterprise EPM financial review cycle, rebuilt as a hierarchical multi-agent system.
Before
4 days
With agents
30 min
02
7 days → 1 day
Logistics reconciliation across previously disconnected supply-chain systems.
Before
7 days
With agents
1 day
03
−50% triage latency
Security alert triage in an air-gapped, on-prem deployment, holding 99.9% uptime.
Before
baseline
With agents
half the time
04
5× faster to care
Clinic voicemail triage that gets urgent patients to a nurse first.
Before
baseline
With agents
5× faster
02 · HOW I WORK
Discover, decide, ship.
Then make it boring.
Every engagement runs the same loop, from technical discovery to production rollout and the product changes that follow.
Find the real problem with the customer's team
discovery · architecture
Put rules where rules belong, and language where language belongs
deterministic engines · multi-agent
Keep a human on anything consequential
human-in-the-loop
Evaluate before release, and keep evaluating after
LLM-as-a-judge · tests
Deploy reproducibly, even air-gapped
cloud tenants · on-prem
Turn what the field teaches into the platform
field-to-product loop
03 · RANGE
Finance. Healthcare. Security. Logistics.
Azure tenants to air-gapped networks.
If it has to work inside a real company, under real constraints, I've probably shipped it there.
10
CUSTOMER PROJECTS
5
AS TECHNICAL LEAD
20+
AGENTS IN PRODUCTION
2
PATENTS PENDING
04 · FEATURED
Work in production.
Platform components I architected, and deployments described by industry.
Judge Council
Multi-stage LLM-as-a-Judge evaluation, before release and on live traffic.
ContextForge
Declarative RAG/CAG context engineering with self-correcting agents.
Financial review agents
A supervisor delegating to retrieval, anomaly, and narrative sub-agents.
Alert triage
ReAct-style LangGraph + MCP on-prem. I led a team of three.
Slack & Teams agents
Query and reconcile EPM, ERP, and accounting data in chat.
FogBrain
Private LLM inference on a Raspberry Pi edge cluster.
2025 → now
Founding FDE · MoolAI
Engineer #3. I build the agent platform (two provisional patents) and lead delivery for enterprise customers from scoping through go-live.
2023 → 2025
M.S. Analytics · SF State
Graduate research on FogBrain, private LLM inference on a Raspberry Pi edge cluster. Runner-up in the CSU Business Analytics Competition 2024.
2021 → 2023
Data Engineer · LTIMindtree
Streaming and lakehouse pipelines on Databricks, Kafka, and Azure DevOps. Processed 2TB+ of IoT data a day and cut ETL from 7 days to 2.
2017 → 2021
B.E. Computer Science · VTU
Visvesvaraya Technological University, Bengaluru. Co-authored a published paper on currency authentication with image processing and k-NN.
