Amritpal Singh - Big Data Developer

Overview

Frontend app engineer (senior-level) focused on building robust browser single-page UIs that orchestrate complex asynchronous backend interactions and file operations. The strongest proven skill is resilient async state and sync logic as implemented in psff-torrents-ui/index.html (applyTorrentSync, scheduleGcsSyncPolling, processPendingTopLevelRenames). There is limited public evidence of automated test coverage, modern JS framework usage, server-side rendering, or cross-team architecture leadership.

Phone

Timeline

Senior Software / Data Engineer · Senior

GridX · Full-Time

Aug 2015 to Present 11 Years 3 Months Delhi Partially remote

Built and operated distributed data pipelines using Python and PySpark, including SparkSQL-based transformations and validation for large analytical workloads. Designed Airflow orchestration for parallel processing and implemented reusable OOP Python components with logging, retries, and production debugging. Integrated multiple databases and implemented REST/SOAP clients and backend services with Django/Flask patterns and OAuth2/JWT security. Added LLM/RAG-based enrichment prototypes and delivered CI/CD container deployments to Kubernetes using AWS and DevOps tooling.

AI Agents

Agentic Workflows

SQL

PostgreSQL

Snowflake

MySQL

LangGraph

Rest API

Selenium

LangChain

Spark

Airflow

Flask

Amazon Redshift

Rancher

GitHub Actions

GitHub

Amazon S3

SOAP

Beautiful Soup

CI/CD

ArgoCD

Pandas

NumPy

Django

Git

AWS

Docker

Kubernetes

Jenkins

Punjab Technical University (PTU)

Bachelor's Degree · Computer Science

2011–2015 Jalandhar, Punjab

Projects

Energy Data Quality & Pipeline Validation

Aug 2015 to Present 11 Years 3 Months

Created a rule-driven validation framework for forecast-versus-actual, settlement-versus-forecast, interval completeness, and data consistency. Embedded quality controls and alerting into scheduled Airflow workflows to improve ingestion reliability, observability, and recoverability.

Python

SQL

Airflow

Trade Reconciliation & Exposure Analytics Platform

Aug 2015 to Present 11 Years 3 Months

Developed a configuration-driven ETL and analytics platform for trade reconciliation, exposure analysis, and Net Open Position reporting, enabling market onboarding through configuration changes.

Python

SQL

Wholesale Settlement & Allocation Engine

Aug 2015 to Present 11 Years 3 Months

Built a configurable wholesale settlement and allocation engine using Python, SQL, PySpark, and Pandas for invoices, market-specific charges, and meter-level allocation.

Python

SQL

pySpark

Pandas

Energy Market Data Collection & Normalization

Aug 2015 to Present 11 Years 3 Months

Automated extraction, validation, and normalization of generation and trading data from Shell Energy Japan power plants using Python, Selenium, BeautifulSoup, Requests, Pandas, and Airflow.

Python

Selenium

Airflow

Pandas

Beautiful Soup

PowerUp

Sep 2017 to Nov 2017 2 Months

Built a mobile app using React.

React.js

PowerEASE

Aug 2015 to Aug 2017 2 Years

Developed a working Android app using Java APIs and a database.

Java

Android SDK

Technical skills

Languages

6

Java

SQL

Scala

Go

Node JS

Python

4

Django

Flask

Beautiful Soup

pySpark

DevOps

8

GCP

Amazon S3

AWS

Kubernetes

Docker

Terraform

CI/CD

AWS CDK

Databases

8

PostgreSQL

Google BigQuery

Snowflake

Redis

MySQL

Cassandra

Databricks

Delta Lake

Other

11

Spark

Azure Data Factory

Axios

Express

Dimensional Modeling

LLM

FastAPI

Dotenv

Frontend

RAG

ETL/ELT

Senior Backend Developer Confidence: High API Engineer

Backend AI/LLM engineer (senior-level) focused on building OpenAI-compatible adapter services with real-time streaming and safety controls. The strongest proven skill is designing an LLM adapter and safe input handling as shown by the provider mapping, SSRF prevention and SSE streaming in src/providers/vertex-gemini.js and src/utils/sse.js. There is limited evidence of production-grade secrets management, RBAC, distributed scaling patterns or database schema evolution in the public code.

API Design

5/10

How well APIs are designed

Clear OpenAI-compatible API surface with SSE streaming and consistent response shaping; pragmatic error responses exist but there is no explicit versioning strategy, idempotency key handling, or advanced API governance.

Evidence

itzAmrit/openai-adapter/src/routes/responses.js: handleResponses (routes, SSE negotiation and error responses) itzAmrit/openai-adapter/src/providers/vertex-gemini.js: mapGenerateContentToResponses (OpenAI Responses shaping) itzAmrit/openai-adapter/test/openai-compat.test.js: tests validating response shapes and fallback model ordering

Data Layer & Database

2/10

Working with databases

A minimal disk-backed function-call cache is implemented with pruning and persistence, but there is no DB schema evolution, migrations, transaction boundaries, or advanced data-layer design.

Evidence

itzAmrit/openai-adapter/src/utils/function-cache.js: CACHE_FILE_PATH based cache, loadCacheFromDisk, saveCacheToDisk

Scalability & Performance

4/10

Handling load and speed

Good pragmatic performance controls: request timeouts, payload size limits, model fallback and transient-retry logic, and streaming output; lacks sophisticated backoff/jitter, measured tuning, or multi-process scaling artifacts.

Evidence

itzAmrit/openai-adapter/src/providers/vertex-gemini.js: shouldRetryTransientUpstreamError and retry loop with model fallback itzAmrit/openai-adapter/src/providers/vertex-gemini.js: MAX_INLINE_IMAGE_BYTES and axios download options (maxContentLength, timeout) itzAmrit/openai-adapter/src/utils/sse.js: streamResponsePayload (SSE streaming implementation)

System Architecture

4/10

Overall system structure

Logical module separation (providers, routes, utils, tests) and focused single-process adapter design; the architecture fits an adapter service but there is no evidence of multi-service orchestration or cross-service contracts.

Evidence

itzAmrit/openai-adapter/src/providers/vertex-gemini.js: provider layer encapsulating request mapping and upstream calls itzAmrit/openai-adapter/src/routes/responses.js: router wiring and streaming vs non-stream branching itzAmrit/openai-adapter/src/utils/function-cache.js: cross-cutting cache utility used by provider

Security & Auth

6/10

Protecting data and access

Strong input validation and SSRF protection for image inputs, HTTPS enforcement, MIME/type checks, and a token refresh flow for auth; broader authn/authz, secrets lifecycle, and dependency audit are not visible in the code analyzed.

Evidence

itzAmrit/openai-adapter/src/providers/vertex-gemini.js: isPrivateAddress, dns.lookup and downloadRemoteImage (SSRF prevention and private address checks) itzAmrit/openai-adapter/src/providers/vertex-gemini.js: normalizeImageUrl and protocol checks (HTTPS only) and MIME validation itzAmrit/openai-adapter/src/providers/vertex-gemini.js: getAccessToken usage and 401 refresh handling

Reliability & Observability

4/10

Stability and monitoring

Reliability is addressed via explicit timeouts, retry detection, fallback model logic, SSE failure events and tests; lacks exponential backoff with jitter, structured logging, metrics/health endpoints and graceful shutdown hooks.

Evidence

itzAmrit/openai-adapter/src/providers/vertex-gemini.js: callVertexGenerateContent timeout and shouldRetryTransientUpstreamError with sleep-based retry itzAmrit/openai-adapter/src/utils/sse.js: buildFailureEvents and streamResponsePayload (SSE failure handling and event stream lifecycle) itzAmrit/openai-adapter/test/openai-compat.test.js: tests for failure event emission and caching behavior

Expertise

Backend AI & LLM· Senior

Node.js· Middle

Microservices & API Architecture· Middle

Messaging & Real-time· Middle

Industries

Artificial Intelligence· Senior

Technologies

Node JS· Senior

Flask· since 2015

Express

FastAPI

Django· since 2015

Axios

Dotenv

Recommendations

Develop LLM adapter and inference-proxy services that require safe input validation, model fallback and SSE streaming behavior.
Build API integrations and middleware for real-time developer tools or coding assistants where input sanitization and tool-call mapping matter.
Implement production hardening for single-process services: add structured logging, metrics/health endpoints, exponential backoff with jitter and secrets lifecycle management.
Extend disk-backed caches to durable storage with migration paths and transactional guarantees when more than ephemeral state is required.

Repositories

The developer's experience in this domain has been verified based on AI analysis of the following repositories:

build-tools-v3 openai-adapter

Senior Frontend Developer Confidence: High App Engineer

Frontend app engineer (senior-level) focused on building robust browser single-page UIs that orchestrate complex asynchronous backend interactions and file operations. The strongest proven skill is resilient async state and sync logic as implemented in psff-torrents-ui/index.html (applyTorrentSync, scheduleGcsSyncPolling, processPendingTopLevelRenames). There is limited public evidence of automated test coverage, modern JS framework usage, server-side rendering, or cross-team architecture leadership.

UI Component Architecture

5/10

How interface parts are built

Modular, function-driven UI rendering with clear separation between state, DOM references, and render routines, but not componentized into reusable framework components.

Evidence

psff-torrents-ui/index.html: renderTable, renderTorrentRow, renderManageModal functions

psff-remote-file-explorer/index.html: renderEntries, renderBrowser functions

Responsive & Cross-browser

5/10

Works on all screens and browsers

Responsive layouts implemented with CSS grid and media queries and thoughtful mobile fallbacks; no explicit RTL/i18n framework but pragmatic responsive handling is present.

Evidence

psff-torrents-ui/index.html: multiple @media (max-width: ...) blocks adapting grid/table layout

psff-remote-file-explorer/index.html: media queries that switch grid columns and toolbar layout

Performance Optimization

6/10

Speed of the interface

Measured and pragmatic optimizations: sliding-window speed sampling, polling cadence control, caching and size-calculation batching, and tooltip caching to reduce backend load.

Evidence

psff-torrents-ui/index.html: updateGcsSyncLiveMetrics implements sampling window and speed calculation; scheduleGcsSyncPolling manages polling cadence

psff-remote-file-explorer/index.html: calculateDirectorySize with caching and queueDirectorySize to avoid duplicate work

psff-torrents-ui/index.html: peer tooltip caching in state.peerTooltip.cache with TTL

Accessibility & Semantics

5/10

Usable for everyone

Good semantic markup and some ARIA usage (live regions, dialog roles, aria-label on progress) and keyboard handlers; not a complete a11y system but clear attention to accessibility in places.

Evidence

psff-torrents-ui/index.html: restoreGate has role="status" and aria-live="polite", modal-backdrop uses role="dialog" and aria-modal="true", progress elements include aria-labels

psff-torrents-ui/index.html: keyboard handling for Enter on form inputs and modal controls

State Management & Data Flow

7/10

Managing data in the app

Strong client-side state discipline and resilient server-state sync patterns: incremental sync with rid, in-flight guards, optimistic updates, persistent retry queues, cancellation for uploads, and careful error handling.

Evidence

psff-torrents-ui/index.html: state object, syncTorrents and applyTorrentSync with rid-based incremental updates, processPendingTopLevelRenames persisted in localStorage

psff-remote-file-explorer/index.html: uploadFileWithProgress uses XMLHttpRequest with abort support and upload cancellation state; calculateDirectorySize uses generation tokens and promise caching to prevent races

UX & Visual Polish

6/10

Look and feel quality

Comprehensive UX with loading, empty and error states, progress/skeleton UI, ETA and speed displays, and meaningful user feedback across flows.

Evidence

psff-torrents-ui/index.html: restore UI, sync-progress-card, renderYtDlpJobs with download and postprocess progress and error messages

psff-remote-file-explorer/index.html: upload status card, progress bar, and detailed upload error/ETA handling

Verified artifacts

anomalyco/opencode· ★ 212.8K

Expertise

PWA & Web APIs· Middle

Web Performance & Optimization· Middle

Technologies

Frontend

GCP· since 2015

Recommendations

Build operator-facing single-page admin consoles that require resilient polling, incremental sync and optimistic actions against complex backend APIs.
Implement browser-based file management and upload flows that need progress, cancellation, and server-side transfer orchestration.
Lead development of small-framework or vanilla-JS control panels where low-dependency, highly inspectable frontends are required.

Repositories

The developer's experience in this domain has been verified based on AI analysis of the following repositories:

psff-torrents-ui build-tools-v3 psff-remote-file-explorer

Job preferences

Work permits: India
Citizenship: India
Salary: from $80k per year
Employment type: Full-Time