Subhayu Kumar Bala profile photo 1 on Tal

Subhayu Kumar Bala

Data Engineer @ Loop AI

Greater Bengaluru Area

4+ years of work experience

Loop AI
Jan 20269 mosPresent

Bengaluru, Karnataka, India

Platform engineer on Loop's core data team, owning ingestion, onboarding, and observability for the food-delivery analytics platform serving major restaurant chains. Key Contributions: • Built Loop Pulse, a full-stack internal observability platform (38 views, 27 API modules) used daily by 6+ stakeholders for ingestion monitoring, SLA tracking, and incident triage across delivery providers. • Bootstrapped TruROI Ops, a plan-generation platform for the finance team (90 PRs in 45 days). Replaced a CDC pipeline with direct Postgres reads, cutting data lag from 13.7 hours to near-zero and query costs by 16x. • Owned numerous production incidents with explicit root causes, including a $14k BigQuery cost runaway response and authored cross-repo cost-safety policy, coordinating recovery campaigns and resolving provider-level failures across multiple 3P platforms. • Optimized BigQuery query path from 342 GB to under 6 GB via partition pruning and incremental dbt modeling, cutting scan costs by 57x.
FiftyFive Technologies
Jun 20223 yrs 8 mosJan 2026

Delivered end-to-end data and AI solutions for 10+ enterprise clients across healthcare, finance, logistics, and tech. Conducted 100+ technical interviews, led client scoping calls, and mentored teammates. Key engagements: • SBGC Group (Retail Data Platform): Owned a multi-region Azure/Databricks platform integrating SAP, Shopify, Klaviyo, and GA4. Unblocked Advent International due diligence by reverse-engineering Shopify's revenue logic to cut variance from ~10% to <0.22%. Root-caused a data explosion bug, slashing volume by 99%. • CV Advisors (Performance Engineering): Replaced a 1900-line SQL stored procedure with vectorized DuckDB/Pandas logic, reducing runtime from 27.5 hours to under 5 seconds on 80M+ records across 150 clients. • QxLab (LLM Infrastructure): Led a 4-person team fine-tuning Llama/Mistral on H100 clusters to >95% accuracy in a production agentic system. Architected a TB-scale dedup pipeline (MinHash LSH + FAISS), later open-sourced as DatasetPipeline (9.7k+ downloads). • AGY Logistics (IoT & Automation): Sole architect of a GDP-compliant cold chain system on GCP. Automated workflows previously requiring 10-15 US-based ops staff. Reduced false positive alerts by 70%. • Johnson & Johnson (Healthcare CDP): Architected Treasure Data pipelines for regulated APAC markets (JP/ANZ). Built a 12-scenario SQL truth table for hierarchical consent resolution, ensuring GDPR/PII compliance. • LoopKitchen (Data Ingestion): Built core data architecture on GCP (BigQuery, FastAPI) enabling a $6M Series A. Migrated legacy scrapers to official API integrations for UberEats, DoorDash, and Grubhub. • Prospexs (GenAI Platform): Built an AI outreach platform backed by a 700M+ contact database with a 4-entity personalization engine (OpenAI/Perplexity), improving response rates by 45%.
FiftyFive Technologies
Jan 20225 mosMay 2022

Gained experience with production infrastructure, backend best practices, and client-facing delivery. Engagements: • Eningo: Developed a cloud-based geospatial platform for cable network automation. Migrated legacy data from Postgres to MongoDB for enhanced scalability. • NIBE: Managed and established Azure DevOps CI/CD pipelines for critical services supporting 50,000+ daily users.

Subhayu’s AI Stack

claude code
claude code
codex
codex
mcp
mcp
Agno
Agno
Browser Use
Browser Use
Wispr Flow
Wispr Flow

Education

Calcutta Institute of Engineering and Management

Bachelor of Technology - BTech, Information Technology

Calcutta Institute of Engineering and Management

20184 yrs2022

Higher Secondary, Science

Amrita Vidyalayam, Durgapur

20162 yrs2018

Join Subhayu and 10k+ other professionals on

Tal Boss

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

Download Tal Boss

Subhayu Kumar Bala on Tal

Headline

Data Engineer @ Loop AI

About

Data and infrastructure engineer specializing in high-performance data systems and production AI/LLM pipelines. Currently at Loop AI, previously shipped solutions for 10+ enterprise clients at FiftyFive Technologies, including Johnson & Johnson, Advent International (SBGC Group), and a McKinsey partner firm. My work sits at the intersection of traditional data engineering and modern AI. I've replaced a 27.5-hour SQL process with a 5-second DuckDB/Pandas engine on 80M+ records, built agentic LLM systems fine-tuned on H100 clusters with >95% accuracy, and architected the data platform behind a startup's $6M Series A. I also write open-source developer tools: 6 Python packages on PyPI with 43,000+ downloads. Published researcher in IET Quantum Communication (quantum computing simulation). Python | SQL | Databricks | BigQuery | DuckDB | Airflow | Kafka | OpenAI | LangChain | AWS | Azure | GCP | Docker | K8s | Terraform

Location

Greater Bengaluru Area

AI stack

claude code

codex

mcp

Agno

Browser Use

Wispr Flow

Work experience

Data Engineer at Loop AI, Jan 2026 - Present · 9 mos, Bengaluru, Karnataka, India, Platform engineer on Loop's core data team, owning ingestion, onboarding, and observability for the food-delivery analytics platform serving major restaurant chains. Key Contributions: • Built Loop Pulse, a full-stack internal observability platform (38 views, 27 API modules) used daily by 6+ stakeholders for ingestion monitoring, SLA tracking, and incident triage across delivery providers. • Bootstrapped TruROI Ops, a plan-generation platform for the finance team (90 PRs in 45 days). Replaced a CDC pipeline with direct Postgres reads, cutting data lag from 13.7 hours to near-zero and query costs by 16x. • Owned numerous production incidents with explicit root causes, including a $14k BigQuery cost runaway response and authored cross-repo cost-safety policy, coordinating recovery campaigns and resolving provider-level failures across multiple 3P platforms. • Optimized BigQuery query path from 342 GB to under 6 GB via partition pruning and incremental dbt modeling, cutting scan costs by 57x.

Data Engineer at FiftyFive Technologies, Jun 2022 - Jan 2026 · 3 yrs 8 mos, Delivered end-to-end data and AI solutions for 10+ enterprise clients across healthcare, finance, logistics, and tech. Conducted 100+ technical interviews, led client scoping calls, and mentored teammates. Key engagements: • SBGC Group (Retail Data Platform): Owned a multi-region Azure/Databricks platform integrating SAP, Shopify, Klaviyo, and GA4. Unblocked Advent International due diligence by reverse-engineering Shopify's revenue logic to cut variance from ~10% to <0.22%. Root-caused a data explosion bug, slashing volume by 99%. • CV Advisors (Performance Engineering): Replaced a 1900-line SQL stored procedure with vectorized DuckDB/Pandas logic, reducing runtime from 27.5 hours to under 5 seconds on 80M+ records across 150 clients. • QxLab (LLM Infrastructure): Led a 4-person team fine-tuning Llama/Mistral on H100 clusters to >95% accuracy in a production agentic system. Architected a TB-scale dedup pipeline (MinHash LSH + FAISS), later open-sourced as DatasetPipeline (9.7k+ downloads). • AGY Logistics (IoT & Automation): Sole architect of a GDP-compliant cold chain system on GCP. Automated workflows previously requiring 10-15 US-based ops staff. Reduced false positive alerts by 70%. • Johnson & Johnson (Healthcare CDP): Architected Treasure Data pipelines for regulated APAC markets (JP/ANZ). Built a 12-scenario SQL truth table for hierarchical consent resolution, ensuring GDPR/PII compliance. • LoopKitchen (Data Ingestion): Built core data architecture on GCP (BigQuery, FastAPI) enabling a $6M Series A. Migrated legacy scrapers to official API integrations for UberEats, DoorDash, and Grubhub. • Prospexs (GenAI Platform): Built an AI outreach platform backed by a 700M+ contact database with a 4-entity personalization engine (OpenAI/Perplexity), improving response rates by 45%.

Software Engineer Intern at FiftyFive Technologies, Jan 2022 - May 2022 · 5 mos, Gained experience with production infrastructure, backend best practices, and client-facing delivery. Engagements: • Eningo: Developed a cloud-based geospatial platform for cable network automation. Migrated legacy data from Postgres to MongoDB for enhanced scalability. • NIBE: Managed and established Azure DevOps CI/CD pipelines for critical services supporting 50,000+ daily users.

Education

Bachelor of Technology - BTech, Information Technology, Calcutta Institute of Engineering and Management

Higher Secondary, Science, Amrita Vidyalayam, Durgapur

Public links

github: https://github.com/subhayu99

medium: https://medium.com/@balasubhayu99

subhayu.in: https://subhayu.in/