Software engineer · full-stack products · applied ML

ShreyanshSharma

I build software products and ML-backed features.

From customer messaging to on-chain flows and verification tools, I focus on products that feel clear to use and reliable underneath.

EmailDirect outreach

LinkedInProfessional profile

ResumeLatest PDF

GitHubCode and repos

Scroll to explore

Real products in productionFull-stack engineeringML-backed featuresBuilt for real useLong Beach, CaliforniaReal products in productionFull-stack engineeringML-backed featuresBuilt for real useLong Beach, CaliforniaReal products in productionFull-stack engineeringML-backed featuresBuilt for real useLong Beach, California

Customer messagingAid distributionCredential verificationAI marketplaceClear product demosCustomer messagingAid distributionCredential verificationAI marketplaceClear product demosCustomer messagingAid distributionCredential verificationAI marketplaceClear product demos

Featured work

RETAIN AI · Founder & Solo Engineer · 2025–Present

RETAIN AI

AI communications infrastructure for customer-facing businesses.

Built and operate live AI communications infrastructure. It handles inbound customer messages for local service businesses with business-specific rules, memory, and AI assistance instead of generic chatbot replies.

Tech stack

Next.jsTypeScriptSupabasePostgreSQLGeminiTwilioVercelPlaywrightVitest

Open platform →

Live platform

Live

retain-ai-eight.vercel.app

Live product window. Open the platform to inspect the full messaging flow.

0+

Scenario variations analyzed

0+

Message exchanges stress-tested

0–0ms

p95 response time under load

0+

Concurrent users stress-tested

<0.4%

Error rate under heavy load

Multi-tenant

Runtime-configurable per business

HYBRID ORCHESTRATION

Deterministic rules + Gemini LLM + memory-driven context. Falls back gracefully at every layer.

MULTI-TENANT RUNTIME

Business identity, tone, FAQs, hours, and workflows are all runtime-configurable. Zero redeploys per tenant.

VOICE — IN PROGRESS

Phone-answering assistant extending the same orchestration pipeline to inbound calls.

Integrations

SupabaseGeminiTwilioPlaywrightVercel

Selected Work

Flagship work

The clearest examples of how I build products people can actually use.

Aid workflow

Live

resilient-aid.vercel.app

Aid protocol

ResilientAid

Moves aid money from donors to beneficiaries and vendors with full on-chain tracking.

Highlights

425 tx/s

Network throughput

0%

Platform fees

<2s

Settlement time

100%

On-chain transparency

4

Stakeholder interfaces

Live

On Polygon Amoy

Tech stack

PolygonSmart contractsAid distributionOffline-capableStablecoin settlement

Features

01

Separate surfaces support admins, donors, beneficiaries, and vendors.

02

Donors can track where funds move instead of sending money into a black box.

03

Beneficiaries receive credits quickly and vendors settle in under two seconds.

04

Every transfer is recorded on-chain with zero platform fees.

Interfaces

Admin PortalOpen Donor PlatformOpen Beneficiary AppOpen Vendor POSOpen

Open →

Marketplace demo

Live

artisian.vercel.app

AI marketplace

KalaAI

Helps shoppers discover handmade goods while AI identifies the likely cultural origin from product photos.

Highlights

120+

Cultural detection categories

5

Product surfaces

AI

Origin detection from imagery

Live

Deployed on Vercel

Specs

Home

Shorts Feed

Marketplace

Trends

Account

Tech stack

AIComputer VisionFull-stackMarketplaceNext.jsVercel

Features

01

A swipeable feed surfaces artisans, products, and short stories quickly.

02

Buyers can browse listings, trends, and seller pages in one place.

03

Image analysis suggests the craft origin across 120+ categories.

04

The AI result shows up inside the shopping flow instead of as a separate tool.

Open →

Verification flow

Live

veri-degree.vercel.app

Credential verification

VeriDegree

Issues digital credentials that schools can publish once and employers can verify in seconds.

Highlights

Mainnet

Algorand deployment

Soulbound

Academic records

Instant

Employer verification

Private

Selective disclosure

Specs

Mainnet credential protocol

Soulbound academic records

Private verification workflow

Tech stack

AlgorandSoulbound credentialsZero-knowledge privacyAcademic verification

Features

01

Schools issue tamper-resistant records instead of sending PDFs around.

02

Students keep the credential in a wallet they control.

03

Employers can verify status from a simple public check.

04

The flow reduces fraud without adding manual review.

Open →

More Work

More shipped work

Smaller builds that still show range, judgment, and execution.

Form scoring demo

Pose → feedback

Coaching output

live score

Depth82

Alignment77

Tempo88

Computer vision

AI Motion Analysis

Scores exercise form from pose data and turns it into coaching feedback someone can use right away.

Features

01

Tracks 17 body keypoints across video frames.

02

Scores depth, balance, tempo, and alignment.

03

Turns pose output into readable coaching notes.

04

Highlights the mistakes that matter instead of dumping raw keypoints.

Specs

17-point MoveNet pose sequences

PyTorch scoring pipeline

Readable coaching output

Tech stack

PythonPyTorchMoveNetOpenCV

Open →

Classifier review

5 classes

Sample 1Ready

Sample 2Ready

Sample 3Ready

Sample 4Ready

Model output

eval run

Superficial91%

Parabasal82%

Metaplastic88%

Medical ML

Cervical Cell CNN

Classifies Pap smear images across five cell types in a medical imaging workflow that is easier to inspect and evaluate.

Features

01

Trained on 4,000 labeled images across five classes.

02

Preprocessing stays consistent before training and evaluation.

03

Transfer learning speeds up training and improves performance.

04

Class confidence makes model mistakes easier to inspect.

Specs

4,000 Pap smear images

Five-class medical image workflow

Reproducible preprocessing pipeline

Transfer-learning evaluation

Tech stack

PythonTensorFlowTransfer learningMedical imaging

Open →

Route monitor

scan window 240ms

WETH

USDC

ARB

WBTC

DAI

Pairs checked13,240+

Spread filterpass

Slippage guardon

RouteDEX A → B → C

Web3 systems

Flashloan Bot

Scans DeFi markets for arbitrage routes and filters out trades that are too risky or too slow to execute.

Features

01

Checks 13,240+ token pairs for route opportunities.

02

Separates scanning, validation, and execution into clear stages.

03

Filters by spread, gas cost, and slippage before acting.

04

Built for automated decisions instead of manual market watching.

Specs

13,240+ token pairs evaluated

Execution guards for timing and slippage

Separated discovery and execution paths

Tech stack

Solidity / EVMDeFiRoute scanningAutomation

Open →

Stack

Core tools behind the work

Languages, frameworks, infra, and platforms that show up repeatedly across shipped products.

Working palette

One surface, split by function. Languages, frameworks, infra, and platforms that repeatedly show up in shipped work.

InterfaceServicesDataMLDeployment

Languages

Core languages for product logic, automation, services, and data handling.

Frontend

UI tooling for product surfaces that need to feel fast, legible, and deliberate.

Backend / Data

Service behavior, data storage, auth, and the layers behind the interface.

ML / Infra

Model workflows, deployment, and system glue when the product needs more than CRUD.

Tools / Platforms

Deployment, collaboration, auth, payments, and practical platform work around shipping.

Experience

Work experience

The two roles that best show my range: enterprise voice AI work at Rezolve.ai and founder-led product engineering on RETAIN AI.

Rezolve.ai

AI Voice Engineer · Contract·2025

Helped build conversational flow logic for the voice assistant's intent recognition and call handling pipeline

Worked on integrating the voice layer with enterprise ITSM systems for real-time ticket creation and escalation

Contributed during early startup phase before the product scaled to Fortune 500 deployments

Voice AINLPITSM IntegrationConversational AILLM Orchestration

Contact

Let’s talk

If the work here lines up with what your team is building, send a note. Email is best. LinkedIn, GitHub, and my resume are right here.

Emailshreyansh.sharma01@student.csulb.eduOpen LinkedInlinkedin.com/in/sharmasshreyOpen ResumeOpen PDFOpen GitHubgithub.com/shrey1110-dotcomOpen