Pinklesh Parjapati — Senior Software Engineer

Experience

FinTech

Senior Software Engineer

Appreciate, Remote

Sep 2024 – Present
● ACTIVE

Architected a multi-tenant B2B US stock & mutual fund platform from scratch using PostgreSQL schemas and scoped APIs.
Slashed API response times from 1–2s to <300ms (85% improvement) via async programming, Redis caching, and SigNoz observability.
Designed a BFF layer and partner-facing API system; authored full API documentation and led partner onboarding calls.
Built AWS Lambda pipelines for real-time capital gains, SIP scheduling, and automated PDF/CSV tax report exports.
Reduced tenant onboarding time by 80% with self-serve tooling; maintained 99.9% uptime on mission-critical services.

FastAPIFlaskPostgreSQL RedisElasticsearchLambda SQSS3SigNoz

LegalTech

SDE-II (Project Lead)

Legistify Services Pvt. Ltd., Gurgaon

Mar 2021 – Sep 2024

Architected and deployed 4 scalable microservices, boosting data accessibility and reliability by 35%.
Led a monolith → microservices migration improving Litigation Management Tool performance by 50% and efficiency by 40%.
Optimised DB queries and caching, achieving a 40% reduction in DB load and 50% faster query response times.
Built a high-performance job scheduler for large-scale data collection, reducing manual effort by 30%.
Developed a bulk data upload module that cut internal data management effort by 80%.

DjangoFastAPIPostgreSQL RedisElasticsearchSQS CronJobAWS

Content Moderation

Senior Representative — YouTube Trust & Safety

Concentrix, Gurgaon

Jan 2018 – Mar 2021

Audited YouTube accounts and channels, making content moderation decisions on videos, comments, and user sustainability.
Promoted to Senior Representative within 2 years based on consistent quality and decision-making accuracy.
Handled high-volume queues for YouTube client services, maintaining SLA compliance and quality benchmarks.

Content ModerationPolicy EnforcementYouTube Decision Auditing

Renewable Energy

Co-Founder & Managing Director

Future Generation Technology Pvt Ltd, Gurgaon

Nov 2017 – Oct 2019

Co-founded a renewable energy consultancy focused on solar power generation solutions.
Managed end-to-end operations including client acquisition, project delivery, and vendor partnerships.
Led strategic planning and business development for the company's initial growth phase.

EntrepreneurshipBusiness DevelopmentRenewable Energy Project Management

Manufacturing

Engineering Trainee

Flash Electronics India Pvt Ltd, Pune

Oct 2015 – Mar 2016

Completed an engineering internship at a leading manufacturer of electronic and electrical auto components for the automobile industry.
Gained hands-on experience with industrial processes, quality assurance, and production workflows.

ElectronicsAutomotiveManufacturing Quality Assurance

Projects

PROJECT_01

Multi-Tenant B2B Investment Platform

Full backend built from scratch for a multi-broker US stocks and mutual funds platform. Covers database schema design, Lambda functions, async background services, Redis caching, and API gateway integrations.

85%

Latency Reduction

80%

Faster Onboarding

99.9%

Uptime

FastAPIPostgreSQLRedisLambda

↗ View on Play Store

PROJECT_02

Litigation Management System (Legistrak)

Led microservices migration and resolved critical performance bottlenecks. Implemented bulk upload pipelines, permission management, audit logging, and full-text search to power a large-scale LegalTech SaaS platform.

50%

Performance Boost

80%

Ops Effort Cut

Microservices Built

DjangoPostgreSQLElasticsearchAWS

↗ View Legistify

PROJECT_03

Web Crawler — Microservice Architecture

Built a microservice architecture from scratch to automate data collection across diverse government websites. Developed a robust job scheduler, captcha-solving service, and orchestrated a monolithic-to-microservices migration for enhanced scalability.

90%

System Improvement

40%

Faster Data Retrieval

60%

Crawler Efficiency

PythonFastAPICelerySQSRedis

PROJECT_04

Crawler Monitoring System

Created a scraper monitoring system to track daily case data, detect successful scrapes, failures, and code errors. Implemented automated notifications for website changes and code bugs, with an automatic retry feature for failed data retrievals.

90%

Error Reduction

Auto

Retry on Failure

24/7

Real-Time Alerts

PythonDjangoPostgreSQLSlack APICron

System Design

Architecture

Multi-Tenant B2B Platform — HLD

Isolated tenant environments using PostgreSQL schema-per-tenant strategy. API Gateway routes traffic; each tenant's data is scoped at the DB layer, never at the application layer, ensuring zero cross-tenant data leakage.

Client A Client B Client C │ │ │ └────────────┴────────────┘ │ API Gateway ← rate limiting, auth │ FastAPI BFF ← tenant resolution ┌─────┴─────┐ │ │ schema_a schema_b (PostgreSQL) │ │ Redis Redis ← tenant-scoped cache

Event-Driven Flow

Async Investment Pipeline

Buy/sell orders enter via REST, are validated and queued to SQS, then processed by Lambda workers that update portfolios, trigger capital gains calculations, and push live prices via Redis pub/sub.

POST /trade → Validate → SQS Queue → Lambda Worker → PostgreSQL → Redis Pub/Sub → Client WS

SQS Lambda Redis Pub/Sub async idempotency keys

Observability

Monitoring & Uptime Stack

SigNoz distributed tracing instruments every request end-to-end. Custom spans track tenant resolution, cache hits, and DB query time. Alerts fire on p95 latency breaches or error rate spikes — enabling the 99.9% uptime SLA.

SigNoz OpenTelemetry p95 latency alerts error rate tracking span instrumentation

Scalability

Horizontal Scaling Strategy

Stateless FastAPI services behind API Gateway scale horizontally. Redis caches NAVs, stock prices, and portfolio summaries with TTL — reducing DB hits by 40%. Lambda functions handle burst load without provisioning.

Load Balancer ┌──────────┼──────────┐ FastAPI-1 FastAPI-2 FastAPI-3 │ │ │ └─────┬─────┘ │ Redis Cache Lambda burst │ PostgreSQL (RDS)

Class Design

Repository Pattern — Data Layer

Each domain (mutual funds, US stocks, auth) has its own Repository class wrapping raw DB access. Services depend only on the repository interface, keeping business logic decoupled from ORM specifics.

MutualFundService └── depends on → IMutualFundRepository │ ┌─────────────┘ ▼ PGMutualFundRepository ← PostgreSQL impl RedisCachedRepository ← cache decorator │ SQLAlchemy ORM → schema_{tenant}

Auth Design

Fine-Grained Access Control

JWT tokens carry tenant_id + user_id + permission_scopes. A FastAPI dependency resolves the tenant on every request and validates scopes against a Redis-cached permission matrix — zero DB round-trips on hot paths.

JWT → Decode → Redis perm cache → Scope check → Handler

JWT FastAPI Depends RBAC Redis perm cache

Data Model

Schema-Per-Tenant PostgreSQL

Each tenant gets an isolated PostgreSQL schema with identical table structures — portfolio, orders, users, transactions. A connection pool middleware sets search_path = schema_{tenant_id} on checkout, transparent to all services.

public schema └── tenants(id, name, config) schema_tenant_001 ├── users(id, email, kyc_status) ├── portfolio(user_id, asset_type, units) ├── orders(id, user_id, status, amount) └── transactions(id, order_id, timestamp) schema_tenant_002 ├── users … (isolated copy) └── …

Background Jobs

SIP Scheduler & Capital Gains Engine

A CronJob-triggered Lambda scans all active SIP mandates, constructs buy orders, and enqueues them atomically to SQS. Capital gains are computed via a stateless function using FIFO/LIFO lot matching — results cached in Redis per user.

CronJob trigger Lambda SQS atomic enqueue FIFO lot matching idempotency Redis result cache

Pattern

BFF — Backend for Frontend

A dedicated BFF layer aggregates data from multiple microservices (stocks, MF, auth) into a single partner-friendly API response. Partners get a stable contract; internal services can evolve independently. Reduces partner integration friction dramatically.

Partner App │ BFF Layer ← single API surface, versioned ┌───┼───┐ │ │ │ StockSvc MFSvc AuthSvc ← internal µservices

Pattern

Cache-Aside with TTL Tiering

Stock prices (volatile) → 5s TTL. NAVs (daily) → 6h TTL. Portfolio summaries (user-specific) → 2min TTL with tag-based invalidation on trade execution. Achieved 40% DB load reduction without stale data issues.

Redis Cache-Aside TTL tiering tag invalidation 40% DB load ↓

Pattern

Strangler Fig — Monolith Migration

At Legistify, migrated a Django monolith to microservices using Strangler Fig — incrementally extracting bounded contexts (search, permissions, notifications) behind a facade router, with zero downtime and a 50% performance boost.

Monolith → Facade Router ├── /search → SearchSvc ├── /perms → PermSvc ├── /notify → NotifySvc └── /legacy → Monolith (shrinking)

Pattern

Outbox + Idempotency for Reliability

Critical financial writes (order creation, SIP execution) use a transactional outbox pattern — the event is written to an outbox table in the same DB transaction as the business record. A poller publishes to SQS; idempotency keys prevent duplicate processing.

DB txn → orders + outbox → Poller → SQS → Lambda (dedup)

Certifications & Study

Scaler Academy — DSA · SQL · LLD · HLD

Completed structured curriculum covering Data Structures & Algorithms, advanced SQL and database internals, Low-Level Design (SOLID, design patterns, class modelling), and High-Level Design (distributed systems, CAP theorem, consistent hashing, system design interviews).

Data Structures & Algorithms Low-Level Design High-Level Design Databases & SQL SOLID principles Design Patterns CAP theorem Consistent hashing Distributed systems

Certified Credentials

Industry Certifications

Cutshort Certified Python — Advanced (Aug 2023)
View Certificate →

Data Structures & Algorithms in Python (Sep 2020)
Coding Ninjas
View Certificate →

Introduction to Python (Apr 2020)
Coding Ninjas
View Certificate →

Stack

Experience

Projects

Impact

System Design

Contact

In Motion