Paritosh Tripathi profile photo 4 on Tal
Paritosh Tripathi profile photo 1 on Tal
Paritosh Tripathi profile photo 2 on Tal
Paritosh Tripathi profile photo 3 on Tal
Paritosh Tripathi profile photo 4 on Tal
Paritosh Tripathi profile photo 1 on Tal

Paritosh Tripathi

Senior Software Engineer @Sharpsell.ai | Distributed and Backend Systems

Bengaluru

4+ years of work experience

sharpsell.ai
Jun 20263 mosPresent

Bengaluru, Karnataka, India

• Designed and built a new analytics platform by migrating from PostgreSQL to ClickHouse, reducing analytical query latency by ~10× for large-scale workloads. • Implemented Change Data Capture (CDC) using PeerDB on AWS ECS, cutting data ingestion and ETL pipelines from hours to minutes while enabling near real-time analytics. • Deployed and operated ClickHouse on Amazon EKS (Kubernetes), building a scalable, highly available analytics infrastructure. • Integrated Apache Superset with ClickHouse to power internal analytics dashboards, delivering faster reporting and improved decision-making. • Optimized the end-to-end analytics pipeline across ingestion, storage, and visualization to support high-throughput analytical workloads. • Built new DBT system called Atlas replacing current PY-ETL system with minimum effort and standarzing the way we write ETL for realtime analytics.
Parspec
Apr 20251 yr 3 mosJun 2026

Bengaluru, Karnataka, India

Software Engineer working at scale — scaling production data systems to millions of writes, evolving the public API to v2. I take systems through the full curve: design and ADRs → production → scale → observability. Inventory Ingestion Service — saga + distributed locks • Designed a saga-based ingestion pipeline that keeps DynamoDB and OpenSearch consistent via compensating rollback, with DynamoDB conditional-write distributed locks serializing concurrent per-SKU writes. • Scaled ingestion from a 5-manufacturer seed to 3.9M+ writes across 7,800+ manufacturers (25–30% MoM growth) at ~0.3 ms avg latency and p99 under 5 ms, absorbing 400K+ single-day bursts with no duplicate-row corruption. Hermes — Ephemeral Environment Platform • Built a FastAPI control-plane that provisions per-branch ephemeral environments on every PR via AWS ECS Fargate, ALB host-header routing, CodeBuild, ECR, and Route53. • Turned a multi-hour manual ops task into a self-service API, growing monthly active developers 7× (2 → 14) and saving ~270 developer-hours/year. Public Application / REST API • Served 100K+ items/hour at sub-100 ms p99 over PostgreSQL, Aurora, DynamoDB, and OpenSearch. • Set REST conventions and error translation that don't leak DB internals, adopted across the API surface. Nexus & Kronos — Async & Cron Platform • Designed Nexus, a unified async-task service (SQS, ECS, WebSocket progress, DLQ-backed retries, queue-depth autoscaling) that moves long-running work off the request path. • Designed Kronos, a centralized cron system that moves in-process cron jobs out of the primary Django service onto dedicated, monitored ECS tasks. Product Finder 2.0 — Search Ranking • Replaced hard attribute filtering with a config-driven weighted soft-ranking engine (zero-deploy weight tuning, match-strength scoring), lifting search accuracy +23%. • Parallelized OpenSearch page fetching with Go goroutines, cutting query latency from 10 ms to 2 ms.
Anakin (YC S21)
Aug 20222 yrs 5 mosDec 2024

Bengaluru, Karnataka, India

Software Engineer and grew into an owner of foundational, high-throughput data systems — building from scratch under real startup constraints. Generic Compiler Service • Built the first version from scratch — a config-driven service (FastAPI, Factory Pattern) transforming unstructured data into structured datasets at 100 GB/day (peak 200 GB/day, 100+ concurrent tasks). • Engineered dual-queue SQS with 15-day DLQ, ECS autoscaling on queue depth, and multi-sink writes to MongoDB, BigQuery, and S3 with write-ahead logs. • Cut new-client onboarding from 3 days to 4 hours. Event-Driven Data Pipeline • Designed the company's event-driven backbone from the ground up, handling 150M+ events/month at 99.9% uptime. • Used partition-aware sharding, async consumers, and backpressure to absorb 50K-events/min peaks. • Pattern reused by several teams to speed up delivery. Android Device-Orchestration Platform • Sole owner — built the platform from nothing (Python, Go), orchestrating automation across on-prem Pixel pools and Genymotion cloud instances with affinity scheduling. • Integrated Frida runtime injection, network recording, and device lifecycle state tracking. • Ran 100+ connected devices across concurrent automation sessions. Microservice Templates & Platform Tooling • Established reusable microservice templates and domain-driven patterns from scratch that 6 teams adopted as their standard. • Compressed feature development from 2 days to 3 hours and enabled independent scaling of 12+ services. • Resolved 50+ critical production incidents and cut infra cost $3K/month. Observability & Cost Optimization • Built BigQuery + Looker Studio analytics and Grafana monitoring on Kubernetes with automated Slack alerting. • Cut reprocessing overhead 40% and on-call escalations 30%. • Reduced defect-resolution time ~40% through better observability. Engineer of the Quarter, Q3 2024 — for backend throughput gains, production-incident resolution, and mentoring junior engineers.
The One World
Dec 20219 mosAug 2022

Bengaluru, Karnataka, India

First production engineering role — shipped ML and backend services behind a social-media platform. Two consecutive tracks: AI engineering, then backend. Backend • Built a containerized FastAPI backend on Python + PostgreSQL + Docker + AWS, lifting API performance ~30%. • Owned API design, schema modelling, and database management end-to-end alongside senior engineers. Machine learning • Designed a recommendation engine and a rooftop-age detection model. • Deployed Dash/Plotly dashboards on AWS EC2, lifting data accessibility for non-technical stakeholders ~25%. Stack: Python · FastAPI · Django · PostgreSQL · MongoDB · Docker · AWS (EC2) · PyTorch · TensorFlow · Plotly/Dash · Linux
GitHub
@paritoshtripathi935

Paritosh’s AI Stack

claude code
claude code
codex
codex
mcp
mcp
opencode
opencode
Wispr Flow
Wispr Flow
Zed
Zed

Education

Bennett University

Bachelor of Technology - BTech, Computer Science

Bennett University

Join Paritosh and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Paritosh Tripathi on Tal

Headline

Senior Software Engineer @Sharpsell.ai | Distributed and Backend Systems

About

Graduated with a Bachelor of Technology in Computer Science from Bennett University, with expertise in backend development and distributed systems. Currently a Senior Software Engineer at Sharpsell.ai, where I contributed to designing and deploying a new analytics platform. This included migrating databases to ClickHouse for 10× faster query performance and implementing near real-time data ingestion pipelines with PeerDB on AWS ECS. Previous roles included scaling production data systems at Parspec and developing high-throughput, event-driven data pipelines at Anakin (YC). Proficient in Python, Go, and ClickHouse, with experience in building scalable architectures and optimizing analytics pipelines. Committed to improving system performance and enabling data-driven decision-making. mailto: paritosh.tripathi.work@gmail.com visit: https://paritosh.online

Location

Bengaluru

AI stack

claude code

codex

mcp

opencode

Wispr Flow

Zed

Work experience

Senior Software Engineer at sharpsell.ai, Jun 2026 - Present · 3 mos, Bengaluru, Karnataka, India, • Designed and built a new analytics platform by migrating from PostgreSQL to ClickHouse, reducing analytical query latency by ~10× for large-scale workloads. • Implemented Change Data Capture (CDC) using PeerDB on AWS ECS, cutting data ingestion and ETL pipelines from hours to minutes while enabling near real-time analytics. • Deployed and operated ClickHouse on Amazon EKS (Kubernetes), building a scalable, highly available analytics infrastructure. • Integrated Apache Superset with ClickHouse to power internal analytics dashboards, delivering faster reporting and improved decision-making. • Optimized the end-to-end analytics pipeline across ingestion, storage, and visualization to support high-throughput analytical workloads. • Built new DBT system called Atlas replacing current PY-ETL system with minimum effort and standarzing the way we write ETL for realtime analytics.

Software Engineer II at Parspec, Apr 2025 - Jun 2026 · 1 yr 3 mos, Bengaluru, Karnataka, India, Software Engineer working at scale — scaling production data systems to millions of writes, evolving the public API to v2. I take systems through the full curve: design and ADRs → production → scale → observability. Inventory Ingestion Service — saga + distributed locks • Designed a saga-based ingestion pipeline that keeps DynamoDB and OpenSearch consistent via compensating rollback, with DynamoDB conditional-write distributed locks serializing concurrent per-SKU writes. • Scaled ingestion from a 5-manufacturer seed to 3.9M+ writes across 7,800+ manufacturers (25–30% MoM growth) at ~0.3 ms avg latency and p99 under 5 ms, absorbing 400K+ single-day bursts with no duplicate-row corruption. Hermes — Ephemeral Environment Platform • Built a FastAPI control-plane that provisions per-branch ephemeral environments on every PR via AWS ECS Fargate, ALB host-header routing, CodeBuild, ECR, and Route53. • Turned a multi-hour manual ops task into a self-service API, growing monthly active developers 7× (2 → 14) and saving ~270 developer-hours/year. Public Application / REST API • Served 100K+ items/hour at sub-100 ms p99 over PostgreSQL, Aurora, DynamoDB, and OpenSearch. • Set REST conventions and error translation that don't leak DB internals, adopted across the API surface. Nexus & Kronos — Async & Cron Platform • Designed Nexus, a unified async-task service (SQS, ECS, WebSocket progress, DLQ-backed retries, queue-depth autoscaling) that moves long-running work off the request path. • Designed Kronos, a centralized cron system that moves in-process cron jobs out of the primary Django service onto dedicated, monitored ECS tasks. Product Finder 2.0 — Search Ranking • Replaced hard attribute filtering with a config-driven weighted soft-ranking engine (zero-deploy weight tuning, match-strength scoring), lifting search accuracy +23%. • Parallelized OpenSearch page fetching with Go goroutines, cutting query latency from 10 ms to 2 ms.

Software Engineer I at Anakin (YC S21), Aug 2022 - Dec 2024 · 2 yrs 5 mos, Bengaluru, Karnataka, India, Software Engineer and grew into an owner of foundational, high-throughput data systems — building from scratch under real startup constraints. Generic Compiler Service • Built the first version from scratch — a config-driven service (FastAPI, Factory Pattern) transforming unstructured data into structured datasets at 100 GB/day (peak 200 GB/day, 100+ concurrent tasks). • Engineered dual-queue SQS with 15-day DLQ, ECS autoscaling on queue depth, and multi-sink writes to MongoDB, BigQuery, and S3 with write-ahead logs. • Cut new-client onboarding from 3 days to 4 hours. Event-Driven Data Pipeline • Designed the company's event-driven backbone from the ground up, handling 150M+ events/month at 99.9% uptime. • Used partition-aware sharding, async consumers, and backpressure to absorb 50K-events/min peaks. • Pattern reused by several teams to speed up delivery. Android Device-Orchestration Platform • Sole owner — built the platform from nothing (Python, Go), orchestrating automation across on-prem Pixel pools and Genymotion cloud instances with affinity scheduling. • Integrated Frida runtime injection, network recording, and device lifecycle state tracking. • Ran 100+ connected devices across concurrent automation sessions. Microservice Templates & Platform Tooling • Established reusable microservice templates and domain-driven patterns from scratch that 6 teams adopted as their standard. • Compressed feature development from 2 days to 3 hours and enabled independent scaling of 12+ services. • Resolved 50+ critical production incidents and cut infra cost $3K/month. Observability & Cost Optimization • Built BigQuery + Looker Studio analytics and Grafana monitoring on Kubernetes with automated Slack alerting. • Cut reprocessing overhead 40% and on-call escalations 30%. • Reduced defect-resolution time ~40% through better observability. Engineer of the Quarter, Q3 2024 — for backend throughput gains, production-incident resolution, and mentoring junior engineers.

AI Engineer · Software Engineer at The One World, Dec 2021 - Aug 2022 · 9 mos, Bengaluru, Karnataka, India, First production engineering role — shipped ML and backend services behind a social-media platform. Two consecutive tracks: AI engineering, then backend. Backend • Built a containerized FastAPI backend on Python + PostgreSQL + Docker + AWS, lifting API performance ~30%. • Owned API design, schema modelling, and database management end-to-end alongside senior engineers. Machine learning • Designed a recommendation engine and a rooftop-age detection model. • Deployed Dash/Plotly dashboards on AWS EC2, lifting data accessibility for non-technical stakeholders ~25%. Stack: Python · FastAPI · Django · PostgreSQL · MongoDB · Docker · AWS (EC2) · PyTorch · TensorFlow · Plotly/Dash · Linux

Education

Bachelor of Technology - BTech, Computer Science, Bennett University

Public links

github: https://github.com/paritoshtripathi935