Anmol Arun profile photo 1 on Tal

Anmol Arun

Full Stack AI Engineer at Impacto Digifin Technologies

Bengaluru

1+ year of work experience

Impacto Digifin Technologies
Sep 20261 moPresent

Bengaluru

Impacto Digifin Technologies
Jun 20263 mosAug 2026

Greater Bengaluru Area

Data & Analytics Platform Engineer building the infrastructure behind a live AML/KYM compliance product for a credit union client. security architecture, upstream OSS debugging, migration validation. SECURITY ENGINEERING Built a custom RBAC system for our BI reporting tool from scratch: a deny-list manager enforcing table-level access at query time, with auto-sync that discovers new tables and prunes stale permissions on a cron. Traced a boot-time bug that silently collapsed 692 permission entries to 0 back to Superset’s built-in role inheritance, then fixed it with targeted exclusion logic. UPSTREAM BUG FIXES Read the source of a third-party MongoDB SQL driver to diagnose two bugs blocking production use: aggregation pipelines silently discarded inside subqueries, and a regex that truncated valid SQL with nested parentheses. Wrote and deployed patches for both, unblocking native MongoDB aggregation inside the reporting tool. Also fixed a SQLAlchemy/Oracle dialect gap via a runtime shim, restoring UAT connectivity without a framework upgrade. MIGRATION VALIDATION Designed an 11-query framework validating a MongoDB-to-PostgreSQL migration across users, banking info replicating source-system dedup logic in SQL. Caught comparability traps mismatched taxonomies, duplicate labels, differing filters that would have caused false failures. Diagnosed a bug with no UI error by querying Superset’s own metadata database directly. INFRASTRUCTURE Administer a multi-container Impacto reporting tool stack (Postgres, Redis, Oracle, MongoDB, Trino). Built a zero-data-loss upgrade procedure across major version bumps. Rejected Trino after reproducing a hard permission conflict, not just reading docs. CLIENT DELIVERY Delivered deployment guidance with access controls and limitations. Built scan alerting, fixing a UTC/IST bug causing false triggers.
Impacto Digifin Technologies
Feb 20265 mosJun 2026

Greater Bengaluru Area

Worked across AML/KYM compliance products, supporting clients, product teams, and operations through a mix of business analysis, data analytics, validation, and process automation. Collaborated with clients to gather requirements, define workflows, write user stories, and translate business needs into product enhancements. Built dashboards and reports in Apache Superset by querying PostgreSQL databases using joins, subqueries, and custom datasets tailored to stakeholder requirements. Led research and onboarding of PEP and SDN data sources across 60+ countries, identifying and validating official government and regulatory sources to strengthen screening coverage. Verified scraped records against source documents, including PDF, XML, and HTML sources, and developed Python-based tools to automate parts of the validation process. Contributed to the design of compliance and lending workflows by refining validation rules, improving data presentation, and helping shape key product interfaces, including PEP profile screens and LOS dashboards. Also supported the delivery of a client-facing website by converting business requirements into a functional solution using Lovable.
Jain (Deemed-to-be University)
Jul 20257 mosJan 2026

Bengaluru, Karnataka, India

Co-authored a reviewed Springer LNNS research paper titled “Voice-Based Emotion Recognition Using CNN and SVM.” Worked on speech signal preprocessing, MFCC feature extraction, normalization, and noise handling. Trained and evaluated an SVM-based classifier using the RAVDESS dataset, achieving ~90–91% accuracy. Focused on robustness, signal variability, and performance evaluation using confusion matrices.
AK APPLE PLY PVT LTD.
Jun 20244 mosSep 2024

Mangaluru

Anmol’s AI Stack

anthropic apis
anthropic apis
antigravity
antigravity
claude code
claude code
claude
claude
codex
codex
cursor
cursor

Education

Jain (Deemed-to-be University)

Bachelor of Technology - BTech, Electrical and Electronics Engineering

Jain (Deemed-to-be University)

20224 yrs2026
presidency school

PUC, Science

presidency school

Join Anmol and 10k+ other professionals on

Tal Boss

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

Download Tal Boss

Anmol Arun on Tal

Headline

Full Stack AI Engineer at Impacto Digifin Technologies

About

i like building things, taking them apart when they break, and being able to explain either one to someone who wasn’t in the room. that mix is what pushed me toward analytics, platform work, and business-facing tech roles rather than picking one lane. my background runs across data, finance, iot, and electrical engineering, so i’m comfortable inside a system and in front of a client. right now i engineer the data platform behind a live aml/kym compliance product for a credit union. that’s meant writing a role-based access control system for impacto’s BI reporting tool from scratch, finding and patching two bugs in a third-party mongodb driver by reading its source, and building an 11-query framework to validate a mongodb-to-postgres migration where the hard part was catching that the two sides were counting different things and would have reported a clean pass either way. i care more about execution than theory. my eee and iot work is where that started load monitoring systems, esp32 and sensor builds, firebase pipelines. things that had to actually run, not just work on paper. the compliance platform work is the same instinct at a different scale: production systems, real clients, no partial credit. on the commercial side i’ve done outreach, customer insight, segmentation, forecasting, and reporting, working directly with founders and small teams. that taught me to own an outcome rather than a task and to move without waiting to be told how. it also means i can sit on a client call and explain what we built and why it matters, which is rarer on technical teams than it should be. tools i work in: python, sql, postgres, mongodb, oracle, superset, docker, pandas, firebase. what i’m looking for: roles that sit between strategy and execution data platform, data science or analytics engineering, product analytics, technical operations. anywhere the problem isn’t clearly defined at the start and someone has to figure out what’s actually wrong.

Location

Bengaluru

AI stack

anthropic apis

antigravity

claude code

claude

codex

cursor

Work experience

Full Stack AI Engineer at Impacto Digifin Technologies, Sep 2026 - Present · 1 mo, Bengaluru

Data Analyst at Impacto Digifin Technologies, Jun 2026 - Aug 2026 · 3 mos, Greater Bengaluru Area, Data & Analytics Platform Engineer building the infrastructure behind a live AML/KYM compliance product for a credit union client. security architecture, upstream OSS debugging, migration validation. SECURITY ENGINEERING Built a custom RBAC system for our BI reporting tool from scratch: a deny-list manager enforcing table-level access at query time, with auto-sync that discovers new tables and prunes stale permissions on a cron. Traced a boot-time bug that silently collapsed 692 permission entries to 0 back to Superset’s built-in role inheritance, then fixed it with targeted exclusion logic. UPSTREAM BUG FIXES Read the source of a third-party MongoDB SQL driver to diagnose two bugs blocking production use: aggregation pipelines silently discarded inside subqueries, and a regex that truncated valid SQL with nested parentheses. Wrote and deployed patches for both, unblocking native MongoDB aggregation inside the reporting tool. Also fixed a SQLAlchemy/Oracle dialect gap via a runtime shim, restoring UAT connectivity without a framework upgrade. MIGRATION VALIDATION Designed an 11-query framework validating a MongoDB-to-PostgreSQL migration across users, banking info replicating source-system dedup logic in SQL. Caught comparability traps mismatched taxonomies, duplicate labels, differing filters that would have caused false failures. Diagnosed a bug with no UI error by querying Superset’s own metadata database directly. INFRASTRUCTURE Administer a multi-container Impacto reporting tool stack (Postgres, Redis, Oracle, MongoDB, Trino). Built a zero-data-loss upgrade procedure across major version bumps. Rejected Trino after reproducing a hard permission conflict, not just reading docs. CLIENT DELIVERY Delivered deployment guidance with access controls and limitations. Built scan alerting, fixing a UTC/IST bug causing false triggers.

Business Analyst at Impacto Digifin Technologies, Feb 2026 - Jun 2026 · 5 mos, Greater Bengaluru Area, Worked across AML/KYM compliance products, supporting clients, product teams, and operations through a mix of business analysis, data analytics, validation, and process automation. Collaborated with clients to gather requirements, define workflows, write user stories, and translate business needs into product enhancements. Built dashboards and reports in Apache Superset by querying PostgreSQL databases using joins, subqueries, and custom datasets tailored to stakeholder requirements. Led research and onboarding of PEP and SDN data sources across 60+ countries, identifying and validating official government and regulatory sources to strengthen screening coverage. Verified scraped records against source documents, including PDF, XML, and HTML sources, and developed Python-based tools to automate parts of the validation process. Contributed to the design of compliance and lending workflows by refining validation rules, improving data presentation, and helping shape key product interfaces, including PEP profile screens and LOS dashboards. Also supported the delivery of a client-facing website by converting business requirements into a functional solution using Lovable.

Undergraduate Researcher at Jain (Deemed-to-be University), Jul 2025 - Jan 2026 · 7 mos, Bengaluru, Karnataka, India, Co-authored a reviewed Springer LNNS research paper titled “Voice-Based Emotion Recognition Using CNN and SVM.” Worked on speech signal preprocessing, MFCC feature extraction, normalization, and noise handling. Trained and evaluated an SVM-based classifier using the RAVDESS dataset, achieving ~90–91% accuracy. Focused on robustness, signal variability, and performance evaluation using confusion matrices.

Financial Analyst at AK APPLE PLY PVT LTD., Jun 2024 - Sep 2024 · 4 mos, Mangaluru

Education

Bachelor of Technology - BTech, Electrical and Electronics Engineering, Jain (Deemed-to-be University)

PUC, Science, presidency school