Note: Dear participants, please make a copy of this doc and delete these instructions.

Do not request edit access on this doc iself.

C4GT DMP - Proposal Template

Durgesh Agrawal
Name

Email ID [email protected]

8329391195
Phone Number

https://github.com/Dmaster007
GitHub ID

944557516963069984
Discord ID

Current occupation Student

(Working Professionals - add current
organization & years of exp)

Indian Institute of technology Bhubaneswar

Education Details
Metallurgy 2025 graduate
(College Name - Degree Name and branch
of engineering or other
course/specialization)

Web development—intermediate, full stack and data

Technical skills with level
structures and algorithms—intermediate​
(Mention tech skills/languages
Docker - novice , react - intermediate, typescript -
known/UI-UX and level -
intermediate, JavaScript - intermediate, Express -
Novice/Intermediate/Expert) intermediate, AWS - novice
Graphql-intermediate, figma-intermediate,
python-intermediate
Title: Unified dashboard for experiments (IDInsight)

Summary
I envision building a Unified Experiment Dashboard that empowers researchers with clear, actionable
insights by leveraging industry‑proven dashboard design principles and a user‑centric development
approach. By choosing the right visualization types—such as line charts with confidence bands for trend
analysis and bar or waterfall charts for segment comparisons—i’ll ensure each chart “tells the story” of
our experiments at a glance. Our approach will prioritize interactivity—filtering, slicing, and
drill-down—to help users explore experiment results on demand, as recommended by data visualization
best practices Technically, i’ll integrate a React/Next.js frontend with performant Python analytics
pipelines, following modular dashboard architecture patterns that support real‑time updates and easy
extensibility. Throughout, i’ll adopt an agile, feedback-driven workflow and apply rigorous dashboard
testing (performance, usability, and accuracy) to deliver a polished, reliable product

Project Detail

1.​ Project Overview:

Please list your understanding of the project (what the requirements are), any issues,
additional support, and probable solutions for them.
​ ​
a.​ Understanding of the project
-​ The unified dashboard will list all experiments with real-time filtering by status
(running/steady/completed), date range, and key metric (Statsig) , and display summary
cards showing top‑line lift and status badges; the enhanced details page will include
time‑series line charts with confidence bands to track metric evolution (Justinmind) and
bar/waterfall charts illustrating effect size across segments (Statsig), alongside a
summary table of key metrics—effect size, confidence intervals, and steady‑state flags
(Wikipedia)—and decision cues based on statistical indicators (p-value thresholds,
sustained lift over N days) to signal when to stop or iterate (Statsig); users will also be
able to dynamically slice data by variant, cohort, or demographic (Klipfolio) and export
raw data or snapshots to CSV/PDF for reporting (Kaltura Knowledge Center); under the
hood, scheduled Airflow or Python scripts will aggregate logs, compute lift, confidence
intervals, and steady‑state status (Monte Carlo Data), with APIs (GET
 /api/experiments , GET /api/experiments/:id/results) providing all
dashboard data.
-​
b.​ Issues that might come up and the support needed from the org (if any)
-​ I anticipate several key challenges and the corresponding support I’ll need: First, data
freshness & pipeline reliability can be compromised by source schema changes or
ingestion failures , so I’ll need access to the experiment log schemas, dedicated
ownership contacts for upstream systems, and support resources that provide clear task
instructions, detailed technology‐usage guidelines, and a framework for constant
feedback and monitoring. Second, the performance vs. freshness trade-off—rendering
near-real-time dashboards can slow queries and degrade UX (Optimizely)—will require
ill-defined SLAs on acceptable data latency and guidance on appropriate refresh
windows. third, statistical threshold configuration—establishing robust criteria for
“steady state” and significance (e.g., p-values, minimum sample sizes)—demands
domain expertise from IDInsight statisticians or experiment owners; And finally, to
prevent UI complexity & usability issues (Justinmind), I’ll need early-stage design
feedback, access to Figma assets, and ongoing collaboration with the design team.
c.​ Solutions
-​ We’ll implement end‑to‑end observability by building pipeline monitoring dashboards
(Acceldata) to track data freshness, volume, and schema consistency and setting up
alerts via Databricks Community so that any SLA breach (e.g., data stale > 24 hrs)
triggers automated fallbacks to cached results; we’ll layer in threshold‑based scheduling
inspired by AoI research (arXiv) to intelligently prioritize updates for the freshest data
without overloading systems; apply cooperative dashboard heuristics (the 39 design
principles from arXiv) with progressive disclosure—showing high‑level KPIs by default
and enabling drill‑downs on demand; Architect the frontend with modular, reusable
React/Next.js components (filter panel, cards, chart wrappers) styled in Tailwind CSS
(Wikipedia), leveraging Chart.js for confidence‑band visualizations and Plotly/D3 for
advanced segment analysis (Statsig); and close the loop with weekly demos to gather
stakeholder feedback, refine our filters and statistical cues, and ensure quality through
automated unit and integration tests (Jest, Cypress) for data accuracy and UI stability.
-​

2.​ Macro Implementation Details with Timelines: Share the macro milestones you will set, along
with timelines (in weeks) for this project.
a.​ Milestone 1
1.​ Setup, Design & Foundation (Weeks 1–2)
2.​ Week 1:​

a.​ Project kickoff, environment provisioning (React/Next.js repo, analytics

workspace)​

b.​ Finalize requirements, API contract​

c.​ Collaborate with design team to produce and review Figma mockups for
both dashboard and detail pages​

3.​ Week 2:​

a.​ Implement core backend scaffolding​

b.​ Configure database/storage schema for aggregated metrics​

c.​ Build shared frontend components: filter panel, card container, chart
wrappers, table shell

b.​ Milestone 2
-​ Core Dashboard & Detail‑Page Features (Weeks 3–4)
Week 3:

●​ Dashboard List:​

○​ Real‑time filtering by status, date range, key metric​

○​ Summary cards displaying top‑line lift and status badges​

●​ APIs Integration: Connect frontend filters to backend endpoints, validate data

flows

Week 4:

●​ Detail Page Visualizations:​

○​ Time‑series line chart with confidence bands​

○​ Bar/waterfall chart for segment effect sizes​

○​ Summary table showing effect size, confidence intervals, and

steady‑state flag​
 ●​ End‑of‑week sync: internal demo to gather quick feedback and refine
components
c.​ Milestone 3
Week 5:​

●​ Export capabilities: CSV data download and PDF snapshot generation​

●​ Implement pipeline monitoring and alerting fallbacks for stale data​

●​ Add dynamic slicing by variant, cohort, and demographic​

Week 6:​

●​ Automated testing: Jest unit tests, Cypress end‑to‑end flows, data validation
checks​

●​ Documentation: README, API docs, “how‑to” guide for exports and filters​

●​ Stakeholder review: demo to IDInsight team, collect feedback, and iterate on

UI/UX and decision‑cue thresholds
-​

Availability

Number of hours available to dedicate to this project per week 15-20 hours

Do you have any other engagements that will require your Yes

time? (projects/internships)

Please note that I’ll be assisting at my father’s shop during daytime hours to help manage the seasonal
demand, but I am on vacation and fully available throughout the specified timeframe.

Personal Information
About Me: I’m Durgesh Agrawal, a software developer and AI enthusiast from Jalna, Maharashtra, and
an IIT Bhubaneswar graduate in Metallurgy & Materials Engineering. With over 800 solved DSA problems
and hands‑on experience in full‑stack web development (Angular, React, Node.js, Django), Python, and
machine learning, I love translating complex challenges into elegant, data‑driven solutions. I’m
passionate about exploring the latest in AI/ML and leveraging technology to build impactful, real‑world
applications.

What is your motivation to apply for this project? Answer briefly in 5-10 lines.
I’m motivated to apply for this project because it perfectly aligns with my passion for transforming
complex data into clear, actionable insights through full‑stack development. Building a unified
experiment dashboard will let me deepen my expertise in React/Next.js, analytics pipelines, and
advanced visualizations while collaborating with the talented IDInsight and Code4GovTech teams—and
learning from experienced mentors. I’m excited by the opportunity to empower public‑sector
researchers to make confident, data‑driven decisions, delivering a polished, user‑centric interface that
drives real‑world impact and furthers my growth as a developer.

Please mention if you have solved any issues/tickets for this or other C4GT projects: (Optional)

Link to to Issue Resolution description in short Link to pull request

Add more rows if required

Previous experience/open source projects (Optional):

In this section you can mention your relevant work experience/projects (not just limited to
open-source). You should mention experiences in this section if any with the relevant tech stack of
the project (for product usability & design projects, the design software you used; like Figma; can
be mentioned)

Project Name Project Description Links (if any)

Add more rows if required