Vietstock Arena
Champion
National student stock market competition, 2023
Evidence Engine - From Noise to Signal
Khiem PhamPham Duc Khiem
I don't predict the future. I structure uncertainty.
I connect finance, AI evaluation, quantitative research, and business execution into workflows that turn market noise, model failures, and complex data into auditable decisions.
CFA Level IVietstock Arena 2023 ChampionFinstock, Inc. FounderAI Red TeamingLLM EvaluationQuantitative Finance ResearchPublic Evidence Trail
Active lens: Recruiter Mode
Highlights finance, business analysis, AI QA, stakeholder communication, and public verification signals.
SIGNAL ROOM / FACE SCAN
input: raw market data + filings + model outputs + research notes
process: structure -> test -> reason -> validate
output: evidence-backed decision
Competition Scale
3,200
Students from 200+ universities and colleges
Public Profiles
4+
GitHub, Twine, LinkedIn, Academia / ORCID-style research traces
Research Focus
Quant + AI
Stochastic modeling, ML, market modeling, risk management
Signal Quest
Run Evidence Scan?
Choose a mission and the site will emphasize the sections, evidence tags, and summary language most relevant to your lens.
mission active: Explorer Mode
Explore Full Signal Map
Current scan: balanced exploration across the evidence engine.
Signal Story
From Noise to Signal
The portfolio's core thesis is simple: messy inputs only become useful when the reasoning path survives review.
01
Noise
Market signals, filings, retail data, model responses, unstructured text, and uncertainty.
filingspricingpromptssalesnews
02
Structure
Financial modeling, prompt design, data cleaning, QA frameworks, rubrics, and source mapping.
rubricmodelschemadatasetassumption
03
Reasoning
Quantitative methods, valuation, risk analysis, mathematical proofs, and LLM evaluation.
valuationriskproofrankingaudit
04
Evidence
Investment decisions, benchmark tasks, research papers, portfolio insights, and audited conclusions.
briefscorepaperdecisiontrace
Khiem's edge is not just knowing finance, AI, research, or operations. It is connecting them into systems that make reasoning traceable.
The Four Modes
The Four Modes of Khiem
The same person presents differently depending on who is reviewing the work: recruiter, investor, AI lab, or technical researcher.
Domain Fusion Map
Where My Domains Converge
Finance, AI, research, and operations are not separate chapters of my profile. They are connected layers of the same operating system: turning uncertainty into structured, verifiable decisions.
Signal routing diagram
Evidence-Based Reasoning
domains connected: 8
Career Signal Timeline
Work History as an Evidence Trail
A signal path across finance operations, founder work, AI model evaluation, red teaming, and quantitative research.
Sep 2023 - PresentVietnam
Founder / Corporate Finance Analyst
Finstock, Inc.
Organization context: Finstock, Inc.'s official site describes an investment and trading support company using advisory services, financial analysis, customized strategies, fundamental / technical / quantitative methods, AI integration, and R&D initiatives.
Built a finance and trading support practice around investment research, market analysis, risk context, and client-tailored strategy support.
Why this matters
This role adds a distinct signal to the system: Founder is connected back to evidence, execution, and reviewable reasoning rather than presented as an isolated job title.
FounderCorporate FinanceEquity ResearchAI Analytics
- Architected Finstock, Inc.'s early-stage vision and managed core financial flows during formation.
- Combined fundamental, technical, and quantitative analysis across U.S. equities, warrants, forex, crypto, macro, and industry contexts.
- Built AI-enabled workflows for market updates, risk assessment, portfolio monitoring, NLP extraction, and investment research.
Mar 2026 - Present
Finance Intern
Eurofins
Organization context: Eurofins is publicly described as a global leader in testing and analytical services across food, environment, pharmaceutical, cosmetic, materials, forensics, and life sciences markets.
Supported finance operations across revenue reconciliation, incentive calculations, reporting, CAPEX analysis, valuation models, and P&L / SG&A review.
Why this matters
This role adds a distinct signal to the system: Financial Analysis is connected back to evidence, execution, and reviewable reasoning rather than presented as an isolated job title.
Financial AnalysisRevenue ReconciliationCAPEXP&L
- Performed revenue reconciliation across Vietnam and Indonesia markets.
- Prepared sales reports and management reporting to track revenue and business performance.
- Calculated incentives and commissions using internal policies and sales records.
Oct 2025 - Mar 2026
LLM Agentic / Associate Researcher
Turing
Organization context: Turing describes itself as a research accelerator for frontier AI labs and an enterprise AI partner, with work spanning reasoning, STEM, multimodality, agents, model evaluation, and high-quality data pipelines.
Designed and evaluated LLM reasoning prompts, model outputs, failure modes, and rubric-aligned training data.
Why this matters
This role adds a distinct signal to the system: LLM Evaluation is connected back to evidence, execution, and reviewable reasoning rather than presented as an isolated job title.
LLM EvaluationPrompt SuitesPreference RankingQA
- Designed single-turn and multi-turn prompt suites for correctness, hallucination risk, instruction following, and reasoning quality.
- Performed pairwise comparisons and preference rankings across multiple model completions.
- Documented failure modes including logic errors, unsupported claims, weak reasoning, and task noncompliance.
Oct 2025 - Apr 2026
MOVE Fellow Expert / Project Phoenix
Handshake AI
Organization context: Handshake's public AI fellowship materials describe flexible project-based work where fellows use subject-matter expertise to help train and improve large language models.
Created graduate-level mathematical proof problems and reviewed model reasoning to expose genuine AI weaknesses.
Why this matters
This role adds a distinct signal to the system: Mathematical Proofs is connected back to evidence, execution, and reviewable reasoning rather than presented as an isolated job title.
Mathematical ProofsAI EvaluationBenchmark Design
- Designed original proof tasks requiring graduate-level mathematical reasoning.
- Wrote complete ground-truth proofs and evaluated model responses against them.
- Documented failure cases that revealed real reasoning errors in AI systems.
Nov 2025 - Jun 2026
Finance Research Expert / Math & Finance Expert
AfterQuery
Organization context: AfterQuery Experts publicly presents a remote expert network for AI training work across finance, data analysis, law, research, and other professional domains.
Built finance-focused AI evaluation questions grounded in public filings and designed to separate surface answers from auditable reasoning.
Finance BenchmarksFilingsAccounting TrapsLLM QA
- Designed questions from U.S.-listed company disclosures, 10-K / 10-Q filings, and financial statements.
- Engineered traps around TTM vs annual, segment vs consolidated figures, diluted shares, and classification differences.
- Balanced difficulty so weak models fail while the task remains fair, traceable, and defensible.
Aug 2025 - Oct 2025
Project Lead Quality / Finance Research Reviewer
TELUS Digital
Organization context: TELUS Digital's AI Data Solutions site describes human-powered data annotation, qualified annotators, subject-matter experts, and a large AI Community supporting training datasets.
Led quality workflows for AI training datasets, annotation validation, reviewer alignment, and finance-oriented review.
Dataset QAReviewer CalibrationFinance Review
- Managed distributed reviewer and QA workflows for annotation and validation.
- Implemented spot checks, sampling audits, escalation paths, and calibration examples.
- Coached reviewers to apply labeling and validation standards consistently across edge cases.
Jun 2025 - Sep 2025
GenAI Associate - Red Teaming
Innodata
Organization context: Innodata publicly describes generative AI test and evaluation, LLM quality assurance, safety and risk evaluation, hallucination checks, security vulnerability testing, and human-in-the-loop validation.
Executed adversarial red teaming to identify unsafe behavior, hallucination patterns, bias / toxicity, and security-style model failures.
Red TeamingTrust & SafetyPrompt InjectionFailure Taxonomy
- Built challenge prompt suites mapped to risk categories and recorded reproducible failure cases.
- Documented severity notes and mitigation feedback for remediation workflows.
- Converted vulnerabilities into training-ready prompt-response examples, labels, and edge-case sets.
2024 - 2025
Expert AI Data Trainer / MSc-PhD Math & STEM QA
Invisible Technologies
Organization context: Public sources describe Invisible Technologies as an enterprise AI platform that structures data, builds workflows, evaluates AI performance, and mobilizes human expertise.
Produced and reviewed high-difficulty math and STEM content for LLM training and evaluation.
Advanced MathSTEM QAData ValidationLLM Training
- Developed Master- and PhD-level math prompts with validation notes and defensible gold approaches.
- Evaluated model answers for correctness, reasoning quality, assumptions, and fragile logic.
- Performed spot checks, guideline compliance checks, and consistency audits.
Mar 2026 - May 2026
Finance Expert / Financial Advisory Cases Reviewer
Vetto AI
Organization context: The specific Vetto AI role remains screenshot-supported from the uploaded evidence archive; public platform context was not strong enough to upgrade this claim.
Reviewed complex investment advisory scenarios for realism, completeness, risk context, and consistency of financial reasoning.
Investment AdvisoryFinancial PlanningRisk Review
- Assessed client constraints, risk factors, and recommendation evolution across multi-step scenarios.
- Provided structured feedback to improve financial advisory case quality.
- Reviewed AI-assisted outputs for nuanced financial decision-making beyond generic templates.
Professional profileEvidence image
Oct 2024 - 2025
Finance Project Lead / STEM Expert / Translator
RWS Group / Outlier / Freelance Clients
Organization context: RWS publicly positions itself as a global AI solutions company with data, content, language technology, model support, and real-world cultural expertise; Outlier publicly describes expert-based AI model training work.
Supported finance, language, and STEM evaluation projects requiring technical precision and quality review.
Finance QASTEMTranslationSubject Matter Expertise
- Reviewed finance-oriented AI and language tasks with subject-matter guidance.
- Evaluated STEM content for technical correctness and structured feedback.
- Supported project-level quality control and technical review workflows.
Sep 2024 - Dec 2024
V-Star Internship
vivo
Organization context: vivo publicly describes itself as a technology company focused on smart devices and intelligent services, with global offline sales and after-sales service centers and large-scale smartphone manufacturing capacity.
Translated retail-chain performance signals into recurring summaries and recommendations for regional commercial execution.
Retail AnalyticsSell-throughCommercial Insights
- Analyzed sell-through and sales performance across 10+ key retail chains.
- Identified channel, SKU, and period gaps and supported a reported 12% sell-through improvement.
- Built recurring performance summaries for retail and channel stakeholders.
Company sourceEvidence image
Mar 2022 - Mar 2023
Sales Specialist / Manager
FAHASA
Organization context: FAHASA publicly presents itself as a long-running Vietnamese book distribution and retail enterprise with a large domestic and foreign-language book footprint.
Managed store-level inventory and merchandising decisions using consumer trend signals and operational discipline.
Retail OperationsInventoryMerchandisingRevenue Growth
- Optimized inventory and visual merchandising to maintain reported 100% product availability in high-traffic periods.
- Analyzed consumer trends to adjust displays and support reported 20% year-over-year branch revenue growth.
Finstock, Inc. OS
Not Black-Box Answers - Reviewable Context
Finstock, Inc. is presented as an operating system for structured investment reasoning: founder work, AI-enabled finance workflows, and raw market inputs become reviewable analyst context.
Product thesis
Finstock, Inc. Intelligence Stack
Not black-box answers - reviewable context for investors and analysts who need finance, AI-enabled monitoring, and risk context to converge into usable decisions.
Equity Research
Technical Analysis
Quantitative Modeling
AI-Enabled Market Monitoring
Portfolio Risk Context
TradingView Indicators
Macro / Industry Reports
Investment Advisory Support
Conceptual interface demo
Finstock, Inc. command desk
not live trading data
Selected module
Market Inputs
Collects the context layer: prices, filings, news, macro notes, watchlists, and client constraints before any conclusion is formed.
source mapwatchlistfilings
Reasoning controls
72% conceptual confidencePrice action
Filings
Macro context
Client constraints
risk meter
conceptual placeholder
news sentiment
conceptual placeholder
exposure chips
conceptual placeholder
This is not a live trading tool. It is a conceptual visualization of the reasoning workflow behind Finstock, Inc.'s research and advisory direction.
Finstock, Inc. represents Khiem's builder mindset: not only analyzing markets, but designing systems that connect research, automation, and risk review so people can reason through markets.
AI Lab
Breaking Models to Build Better Ones
This is the model testing chamber: adversarial prompts, finance-grade benchmarks, mathematical rigor, QA discipline, and traceable model failure analysis.
Finance Reasoning Benchmarks
Questions grounded in public filings and statements, engineered around TTM vs annual, segment vs consolidated, share counts, non-recurring items, and auditable assumptions.
Red Teaming & Safety Evaluation
Adversarial testing for hallucination, unsafe behavior, bias / toxicity, prompt injection, instruction failures, reproducible cases, severity notes, and mitigation feedback.
QA & Dataset Quality
Annotation validation, reviewer calibration, spot checks, sampling audits, edge-case handling, disagreement resolution, and scalable quality control.
Mathematical Reasoning / Project Phoenix
Graduate-level proof problems, ground-truth proofs, model response evaluation, and benchmark tasks that expose real reasoning errors.
Model Failure Simulator
Run a pressure test
Domain
Failure type
Finance / numerical inconsistency
confidence: 84%Simulated model answer
The company's revenue increased, so profitability improved.
Failure detected
Revenue growth does not automatically imply profitability improvement. Margin, cost structure, one-off items, and segment mix must be reviewed.
Khiem's evaluator lens
- Check revenue, gross margin, operating income, and net income together.
- Match quarterly, annual, or TTM periods before drawing a trend.
- Separate recurring operations from one-off gains or losses.
Income statement
Segment notes
One-off item disclosure
Period convention
Research Library
Quant Finance as Reviewable Theory
The research section is not decorative. It connects mathematical modeling, machine learning, real-world finance, fintech applications, and risk management.
The Arena
Backtesting Knowledge Under Pressure
Viet Stock Arena becomes the origin story: a place to test market knowledge, compete nationally, and validate an analytical framework.
Champion signal
Champion / First Prize - Viet Stock Arena 2023
Khiem entered Viet Stock Arena not just to compete, but to test strategies, indicators, and market knowledge in a simulated market arena. The public reports describe a national student competition where participants practiced enterprise analysis, stock analysis, and applied knowledge in a simulated stock market environment.
Backtest
Strategy and technical indicators
Compete
National student arena
Validate
First prize / champion
Animated leaderboard
Competition matrix
01Pham Duc Khiem40%+ final narrativeChampion
02Finalist setRound 2 qualifiedTop 40
033,200 students200+ schoolsNational field
Also listed in the uploaded materials: First Runner-up - RongViet Invest 2023 (Investment competition). This claim remains screenshot-supported by the Webport evidence archive unless separately verified by a public source.
Proof of Discipline
A Terminal Grid of Solved Constraints
ProjectEuler+ is framed as more than a coding badge: it is repeated evidence of mathematical persistence, constraint handling, and exactness.
ProjectEuler+ / HackerRank
#1 Vietnam, top 0.01% globally, 200+ perfect solves
A screenshot-supported discipline signal from the uploaded evidence archive; framed as mathematical persistence rather than a public credential claim.
Why this matters
This is less about a coding badge than repeatable mathematical discipline: exact answers, constraint handling, persistence, and the patience needed to debug reasoning under pressure.
Webport archiveEvidence image
Rank
#1 Vietnam
Global
Top 0.01%
Solved
200+ perfect
Evidence Archive
Trust Built as an Indexed Source Library
A Webport-ordered evidence library. Images appear only inside the section they belong to: finance task screenshots stay with AfterQuery, Project Phoenix stays with Handshake AI, Vetto screenshots stay with Vetto AI, and research cards do not borrow work screenshots.
Webport Evidence Manifest
This archive follows the uploaded Webport sequence, using nearby section context and image content to prevent unrelated screenshots from appearing in the wrong modal.
35
Webport sequence
Image positions parsed from the uploaded Webport document flow.
8
Archive groups
Evidence is grouped to mirror the original Webport narrative order.
11
Source-only cards
Research, certificates, and some roles avoid unrelated image assignments.
13
Image-backed cards
Images are shown only inside their matching Webport section galleries.
Identity & Positioning
Organizations & Memberships
Experience Overview
Career & Company Experience
Finance + AI Evaluation Artifacts
Achievements
Certificates & Research
Extra-Curricular
Personalized Signal Summary
Generate My Signal Summary
A rule-based summary built from your selected mission, sections visited, evidence opened, and signal fragments collected. No external API is used.
Explorer Mode / 0 signals collected
Run the summary after exploring a few sections. It will adapt to the active mission and the signals you have collected.
Contact
Let's build systems where finance, AI, and evidence meet.
A professional contact surface for recruiters, clients, AI evaluation teams, and research reviewers.
Collaboration terminal
Available for finance-grade reasoning systems.
Reach out for finance analysis, investment research, AI evaluation, LLM red teaming, quant research support, business analysis, and evidence-backed consulting.
FinanceAI EvaluationQuant ResearchBusiness AnalysisLLM QAInvestment ResearchConsulting
Verified routes