i anuj profile photo 1 on Tal

i anuj

.

Idukki

1+ year of work experience

Stealth Startup
Mar 20251 yr 1 moMar 2026

Chennai, Tamil Nadu, India

Architected and productionized safe horizontal autoscaling logic for long-running AWS Fargate workers usingcustom CloudWatch math expressions, preventing premature worker termination and guaranteeing100% in-flight work completion during scale-in events. Hardened async workflows using EventBridge andDLQ-backed failure isolation.– Built and productionized background processing workers and distributed message queues in a large-scaledata ingestion infrastructure from scratch, enabling ingestion of 1M+ records/day (founder verified).– Re-architected SQS → Lambda into an SQS → Fargate → SQS → Fargate fan-out pipeline for long-runningparallel workloads.– Scaled raw data fetching per API call by 900% (10 → 100) while ensuring rate-limit compliance; slashedwall-clock latency by 75% (80 → 25 mins) and eliminated 60% data loss.– Provisioned and managed cloud infrastructure using AWS CDK (Python); containerized ingestion workers withDocker for ECS Fargate deployment and configured CloudWatch for production observability.– Leveraged DynamoDB Streams for near-real-time Change Data Capture (CDC), exploiting shardordering guarantees to enforce deterministic async processing and eliminate race conditions across pipelines(DynamoDB → Streams → Lambda → Elasticsearch).– Slashed DynamoDB Stream processing latency by up to 90% by tuning the Lambda parallelization factorfrom 1 → 10 while maintaining strict partition-key ordering guarantees.– Helped in a second DynamoDB CDC subflow; re-architected the stream view from NEW IMAGE toNEW AND OLD IMAGES to implement diff-based synchronization logic, safely updating indexed records whenasync services modified existing items.– Enforced system-wide idempotency across two critical areas: first, by tracking unique event ids to guaranteeexactly-once processing and prevent duplicate webhook messages; second, by introducing UUIDv5-baseddeterministic IDs across pipelines to enable safe retries.
Incedo Inc.
Jan 20249 mosSep 2024

Chennai, Tamil Nadu, India

Designed a MySQL relational schema for a Hackathon Management System within a Spring Bootbackend via Hibernate ORM with @ManyToOne and cascading @OneToOne entity relationships to ensure dataintegrity.– Engineered a multi-criteria scoring engine to evaluate submissions, capturing grading data across 5 technicaldimensions into a composite netRating field while managing project statuses via direct database updates.– Optimized submission routing using a fair-allocation algorithm with randomized distribution logic to assignincoming records uniformly across open panelist queues, eliminating review bottlenecks.– Decoupled API payloads from database entities using Data Transfer Objects (DTOs) to parse multi-user JSONinputs, enabling persistence of nested entity structures.– Hardened application security using Spring Security and BCrypt hashing for user credentials; integratedbackend REST APIs with a React.js frontend using the Fetch API.
GitHub
@ianuj42312

i’s AI Stack

antigravity
antigravity
codex
codex
cursor
cursor

Join i and 10k+ other professionals on

Tal Boss

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

Download Tal Boss

i anuj on Tal

Headline

.

Location

Idukki

AI stack

antigravity

codex

cursor

Work experience

backend eng at Stealth Startup, Mar 2025 - Mar 2026 · 1 yr 1 mo, Chennai, Tamil Nadu, India, Architected and productionized safe horizontal autoscaling logic for long-running AWS Fargate workers usingcustom CloudWatch math expressions, preventing premature worker termination and guaranteeing100% in-flight work completion during scale-in events. Hardened async workflows using EventBridge andDLQ-backed failure isolation.– Built and productionized background processing workers and distributed message queues in a large-scaledata ingestion infrastructure from scratch, enabling ingestion of 1M+ records/day (founder verified).– Re-architected SQS → Lambda into an SQS → Fargate → SQS → Fargate fan-out pipeline for long-runningparallel workloads.– Scaled raw data fetching per API call by 900% (10 → 100) while ensuring rate-limit compliance; slashedwall-clock latency by 75% (80 → 25 mins) and eliminated 60% data loss.– Provisioned and managed cloud infrastructure using AWS CDK (Python); containerized ingestion workers withDocker for ECS Fargate deployment and configured CloudWatch for production observability.– Leveraged DynamoDB Streams for near-real-time Change Data Capture (CDC), exploiting shardordering guarantees to enforce deterministic async processing and eliminate race conditions across pipelines(DynamoDB → Streams → Lambda → Elasticsearch).– Slashed DynamoDB Stream processing latency by up to 90% by tuning the Lambda parallelization factorfrom 1 → 10 while maintaining strict partition-key ordering guarantees.– Helped in a second DynamoDB CDC subflow; re-architected the stream view from NEW IMAGE toNEW AND OLD IMAGES to implement diff-based synchronization logic, safely updating indexed records whenasync services modified existing items.– Enforced system-wide idempotency across two critical areas: first, by tracking unique event ids to guaranteeexactly-once processing and prevent duplicate webhook messages; second, by introducing UUIDv5-baseddeterministic IDs across pipelines to enable safe retries.

sde-1 at Incedo Inc., Jan 2024 - Sep 2024 · 9 mos, Chennai, Tamil Nadu, India, Designed a MySQL relational schema for a Hackathon Management System within a Spring Bootbackend via Hibernate ORM with @ManyToOne and cascading @OneToOne entity relationships to ensure dataintegrity.– Engineered a multi-criteria scoring engine to evaluate submissions, capturing grading data across 5 technicaldimensions into a composite netRating field while managing project statuses via direct database updates.– Optimized submission routing using a fair-allocation algorithm with randomized distribution logic to assignincoming records uniformly across open panelist queues, eliminating review bottlenecks.– Decoupled API payloads from database entities using Data Transfer Objects (DTOs) to parse multi-user JSONinputs, enabling persistence of nested entity structures.– Hardened application security using Spring Security and BCrypt hashing for user credentials; integratedbackend REST APIs with a React.js frontend using the Fetch API.

Public links

github: https://github.com/ianuj42312

i anuj - backend eng in Idukki | Tal