Lomash Sharma — Bangalore, IN · Open to hard problems

I build systems
that scale.

Staff Backend Engineer building distributed systems, multi-tenant SaaS, data infrastructure and AI products — founding engineer at three startups, 0 → 1 every time.

View my work → Let's talk lomashs09@gmail.com
0Enterprise tenants
0Data connectors
MillionsRows / day
0Latency reduction
0Years engineering
Lomash Sharma
Lomash Sharma · Bangalore

Trajectory

From zero → scale.

Three startups. Three platforms built from the first commit. Each step widened the blast radius — from services, to platforms, to whole products owned end to end.

Software Engineer Founding Engineer Staff Backend Engineer Founder / Product Builder

09/2019 — 03/2021

Yelo Bank

Software Engineer

Designed a banking platform's microservices architecture from scratch — 12+ Spring Boot services, centralized IAM, and money-transfer rails moving millions of INR at 99% reliability.

03/2021 — 06/2026

Drivetrain AI

Lead Software Engineer · Founding Engineer

First engineer at a Series A FP&A SaaS. Designed the foundational backend architecture and scaled it from zero customers to 400+ enterprise tenants, leading a 4-engineer pod along the way.

07/2026 — Present

SemanticBrain AI

Staff Backend Engineer, APIs · Founding Member

Founding member at an A.T. Kearney subsidiary. Own the social-data ingestion platform end to end and built the agentic retrieval layer 0 → 1.

2026 — Present

Plumefy

Founder & Full-Stack Engineer

A live multi-tenant social publishing product, built and operated solo — proof that the systems thinking ships as product, too.

Impact

Numbers that shipped
to production.

Not benchmarks. Not prototypes. Measured outcomes on systems that enterprises run their finance and data operations on.

0+

Enterprise tenants

Scaled Drivetrain from zero on schema-per-tenant Postgres.

0+

Data connectors

Salesforce, NetSuite, QuickBooks, HubSpot, Xero and more.

Millions

Rows synced daily

Through Spring Boot, Kafka, Temporal, Airflow and DBT.

0%

Latency reduction

Redesigned core data-access and caching paths.

0%

Faster authorization

Led the full authz redesign — and the case to retire the legacy system.

0%

Onboarding cut

Company-wide DBT rollout freed ~35% of engineering bandwidth.

Selected work

Case studies in
systems that carry weight.

CASE 01 — CURRENT

SemanticBrain AI

Staff Backend Engineer, APIsFounding memberA.T. Kearney subsidiary2026 — Present

Turned a metered, rate-limited vendor API into a predictable raw → gold pipeline on GCS across every supported social platform — with restartable runs, tenant isolation, and a hard credit cap that makes overspend impossible. Then built the agentic retrieval layer that answers questions on top of it.

Social APIs

Metered, rate-limited third-party vendor APIs across every supported social platform. Every call costs credits — the system treats quota as a first-class resource. Hover or tap any stage to explore the pipeline.

Restartable runs Tenant isolation Hard credit cap LangGraph PydanticAI GCS raw → gold
CASE 02 — FLAGSHIP

Drivetrain AI

Lead Software EngineerFounding engineerSeries A · AI-native FP&A SaaS2021 — 2026

Five years from first commit to enterprise platform. I designed Drivetrain's foundational backend architecture and grew it into a data platform that moves millions of rows daily for the finance teams of 400+ companies.

0

customers · 2021

0+

enterprise tenants · schema-per-tenant Postgres

150+ connectors

Connector fleet syncing Salesforce, NetSuite, QuickBooks, HubSpot, Xero and 145+ more sources — plus open-source contributions back to the Singer Tap ecosystem. Select any layer to inspect it.

~40% latency reduction

Redesigned core data-access and caching paths in critical Spring Boot services — measured end to end, in production.

Authorization redesign

Persuaded leadership to retire the legacy authz infrastructure; shipped the replacement with ~25% faster requests and cleaner multi-tenant isolation.

Snapshot system

Daily versioned captures of customer datasets — the foundation for historical analytics and time-series pipeline metrics.

Days → seconds

Standardized transformations for the biggest sources cut time-to-first-value for new customers from days to seconds.

~35% bandwidth freed

The DBT rollout moved onboarding work out of engineering — freeing a third of the team for higher-leverage work.

Led a 4-engineer pod

Technical reviews, hiring loop design, weekly 1:1s, and roadmap alignment with Product.

CASE 03 — FOUNDATIONS

Yelo Bank

Software Engineer0.5Bn Finhealth Pvt Ltd2019 — 2021

A banking platform's backbone, designed from scratch: 12+ Spring Boot microservices under Zeebe-based orchestration, moving real money — millions of INR — at 99% reliability.

API layer · client requests
IAMOTP · MFA · RBAC · Cognito
Money transferNEFT · IMPS · ICICI composite
Salary payoutsevent-driven · −50% time
ServiceSpring Boot
ServiceSpring Boot
ServiceSpring Boot
ServiceSpring Boot
+ 5 more12+ total
Zeebe orchestration · 99% reliability
Event bus · async core modules
12+ Spring Boot services 99% reliability ~10% higher txn success −50% payout processing time Millions of INR processed

Founder · Independent product

I don't just design systems. I ship products.

Plumefy is a live multi-tenant social publishing product — built, launched and operated solo, concurrent with a full-time staff role.

Direct Instagram, Facebook, LinkedIn and Threads integrations — real platform APIs, not embeds.

Per-platform analytics and Stripe subscription billing, multi-tenant from day one.

Solo stack: Next.js / TypeScript, Postgres, Supabase, Vercel.

Visit plumefy.com →
plumefy.com/dashboard

plumefy·

Compose

Calendar

Analytics

Channels

Billing

Shipping week 34 — new per-platform analytics just landed. One composer, four networks, zero copy-paste. 🚀

InstagramFacebookLinkedInThreads
Scheduled · Mon 09:30 ISTPublish
SUBSCRIPTION · STRIPEPRO · ACTIVE

12.4k▲

Reach / wk

3.1k▲

Engagement

4

Channels

AI infrastructure

Building AI systems,
not just calling APIs.

Agentic retrieval in production at SemanticBrain: intent detection routes every question across a knowledge graph and a vector store, and every answer comes back grounded and schema-validated. Watch a question move through it.

input

User question

"How did our audience shift last quarter?"

route

Intent detection

LangGraph routes to the right substrate

structured

Knowledge graph

entities & relationships

semantic

Vector search

embedding similarity

retrieve

Grounded retrieval

evidence, not hallucination

validate

Schema validation

PydanticAI — typed, structured output

output

Answer

grounded · cited · validated

LangGraph PydanticAI RAG Knowledge graphs Vector search Agentic retrieval

How I work

i

Systems over features.

ii

Ownership over handoffs.

iii

Measure before optimizing.

iv

Simple architecture beats clever architecture.

v

Build for scale — don't prematurely optimize.

Toolbox

Depth where it matters.

Backend

JavaSpring BootNode.jsREST APIsMicroservicesTypeScriptPython

Data

PostgreSQLBigQueryDBTAirflowTemporalKafkaRedis

Cloud

AWSEC2 · S3 · LambdaCognitoSES · SNS · SSMGCP

AI

LangGraphPydanticAIRAGKnowledge graphsVector search

Architecture

Distributed systemsMulti-tenant SaaSEvent-driven architectureSystem design

Projects

Built on my own time.

Live product · Founder

Plumefy

Multi-tenant social publishing across Instagram, Facebook, LinkedIn and Threads — with per-platform analytics and Stripe subscriptions. Designed, built, deployed and operated by one person.

Next.js · TypeScript · Postgres · Supabase · Vercel · Stripe plumefy.com →
NLP · 2019

Automated Text Summarizer

Extractive summarizer that condenses long documents while preserving contextual meaning.

Python · Flask · AngularJS
Desktop · 2019

GitHub Desktop Client

Cross-platform desktop app replicating core GitHub Desktop functions for repo management and Git operations.

Cross-platform desktop

Signals

Recognition along the way.

2025

Winner — Best Use of AI, Drivetrain Hackathon

AI
—

Winner — AI for Software Testing, Drivetrain Hackathon

AI-assisted test generation & coverage tooling
—

2nd Prize — Drivetrain Hackathon

Backend innovation · team leadership
Ongoing

Open source — Singer Tap ecosystem contributor

One of the largest open-source data-movement projects
2020

Semi-finalist — CodeGladiator

250,000+ participants
—

Gold badges — HackerRank (Python, Problem Solving)

Ranked 360 in Node.js on TechGig

Contact

Have an interesting problem? Let's build it.

Open to conversations around engineering leadership, distributed systems, AI infrastructure, and ambitious products.

lomashs09@gmail.com
Email me → LinkedIn Plumefy

Bangalore, India · Remote-friendly