diff --git a/2023/programme/01_generate_spreadsheet_from_pretalx_data/.gitignore b/2023/programme/01_generate_spreadsheet_from_pretalx_data/.gitignore
new file mode 100644
index 0000000..09f4f9e
--- /dev/null
+++ b/2023/programme/01_generate_spreadsheet_from_pretalx_data/.gitignore
@@ -0,0 +1,4 @@
+config.toml
+env/
+*.pyc
+data/
diff --git a/2023/programme/01_generate_spreadsheet_from_pretalx_data/.python-version b/2023/programme/01_generate_spreadsheet_from_pretalx_data/.python-version
new file mode 100644
index 0000000..2c07333
--- /dev/null
+++ b/2023/programme/01_generate_spreadsheet_from_pretalx_data/.python-version
@@ -0,0 +1 @@
+3.11
diff --git a/2023/programme/01_generate_spreadsheet_from_pretalx_data/01_download_data_from_pretalx.py b/2023/programme/01_generate_spreadsheet_from_pretalx_data/01_download_data_from_pretalx.py
new file mode 100644
index 0000000..867b27f
--- /dev/null
+++ b/2023/programme/01_generate_spreadsheet_from_pretalx_data/01_download_data_from_pretalx.py
@@ -0,0 +1,48 @@
+import json
+from datetime import datetime
+from pathlib import Path
+from pprint import pprint as pp
+
+import requests
+from config import conf
+
+class Config:
+    token = conf["pretalx-token"]
+    event = "europython-2023"
+
+headers = {
+    "Accept": "application/json, text/javascript",
+    "Authorization": f"Token {Config.token}",
+}
+
+base_url = f"https://pretalx.com/api/events/{Config.event}/"
+
+resources = [
+    "submissions",
+    "reviews",
+    "speakers",
+    "answers",
+]
+
+for resource in resources:
+    print("h1: ", resource)
+    url = base_url + f"{resource}"
+
+    res0 = []
+    data = {"next": url}
+    n = 0
+    while url := data["next"]:
+        n += 1
+        print(f"Page {n}")
+        response = requests.get(url, headers=headers)
+        data = response.json()
+        res0 += data["results"]
+
+    timestamp = datetime.now().strftime("%Y%m%d_%H%M")
+    fnames = [
+        f"data/{Config.event}/{resource}_{timestamp}.json",
+        f"data/{Config.event}/{resource}_latest.json",
+    ]
+    for fname in fnames:
+        with open(fname, "w") as fd:
+            json.dump(res0, fd)
diff --git a/2023/programme/01_generate_spreadsheet_from_pretalx_data/02_parse_and_merge.py b/2023/programme/01_generate_spreadsheet_from_pretalx_data/02_parse_and_merge.py
new file mode 100644
index 0000000..c71f91b
--- /dev/null
+++ b/2023/programme/01_generate_spreadsheet_from_pretalx_data/02_parse_and_merge.py
@@ -0,0 +1,386 @@
+"""
+Parse, merge and combine data from speakers, answers, submissions to generate
+the base dataframe that we can use later to merge reviews and community voting
+results.
+"""
+
+import json
+from collections import Counter
+from datetime import datetime as dt
+
+import numpy as np
+import pandas as pd
+
+pd.options.display.max_rows = 50
+pd.set_option("display.max_columns", 500)
+pd.set_option("display.width", 1000)
+
+sub = pd.read_json("./data/europython-2023/submissions_latest.json")
+rev = pd.read_json("./data/europython-2023/reviews_latest.json")
+ans = pd.read_json("./data/europython-2023/answers_latest.json")
+spk = pd.read_json("./data/europython-2023/speakers_latest.json")
+# vot = pd.read_csv("./data/europython-2023/votes-latest.csv")
+
+
+# First simplify the submissions
+# We can drop the speakers from the original table because we will be later
+# joining with the speakers table anyway.
+# We need a separate join with speakers, because we do separate join with
+# answers - some answers are per submission and some are per speaker.
+
+sub["type"] = sub["submission_type"].map(lambda x: x.get("en", "").strip())
+sub["track"] = sub["track"].map(lambda x: isinstance(x, dict) and x.get("en", ""))
+del sub["submission_type"]
+
+# skip withdrawn talks while we're at it.
+sub = sub[sub["state"] == "submitted"]
+
+
+# clean up answers as well
+ans["q_id"] = ans["question"].map(lambda x: x.get("id", ""))
+ans["q"] = ans["question"].map(lambda x: x.get("question", {}).get("en", ""))
+
+# This is a helper for a temporary groupping of questions and answers that we
+# use for both submissions and speaker answers.
+# The details don't matter that much because we later convert those answers to
+# separate columns
+
+
+def _qa_dict(row):
+    q_id_list = row["q_id"]
+    q_list = row["q"]
+    a_list = row["answer"]
+    out = []
+    for i, q, a in zip(q_id_list, q_list, a_list):
+        out.append({"q_id": i, "q": q, "a": a})
+
+    return out
+
+
+def _get_answer(answers, _id):
+    """
+    Helper to extract answers to specific questions, given by question ID from pretalx
+    """
+    q = [x.get("a") for x in answers if x.get("q_id") == _id]
+    return ", ".join(q)
+
+
+# Big Picture
+# ===========
+# 1. We start with prepping the speakers table, and we merge it with answers.
+# 2. Then we merge answers and clean up submissions.
+# 3. Then we merge new cleaned up speakers table with new cleaned up
+# submissions table.
+#
+# This gives us basic table with proposals that we can merge with voting and
+# reviews data.
+
+spk = spk.rename(
+    columns={"code": "speaker_code", "name": "speaker_name", "biography": "bio"}
+)
+
+speakers1 = pd.merge(
+    spk,
+    ans[["q_id", "q", "person", "submission", "answer"]],
+    left_on="speaker_code",
+    right_on="person",
+    how="left",
+)
+
+def implode(df, keep: list[str], agg: list[str]):
+    assert isinstance(keep, list)
+    assert isinstance(agg, list)
+
+    df = df.groupby(keep, group_keys=True, dropna=False).aggregate(
+        {col: lambda x: x.tolist() for col in agg}
+    )
+    df.reset_index(inplace=True)
+    df = df.loc[:, keep + agg]
+
+    df["Q"] = df.apply(lambda row: _qa_dict(row), axis=1)
+    df = df.drop(["q_id", "q", "answer"], axis=1)
+    return df
+
+
+# Temporary copy of submissions
+temp = speakers1[["speaker_code", "submissions"]]
+temp = temp.drop_duplicates("speaker_code")
+temp = temp.set_index("speaker_code")
+
+# Implode the version without submissions
+# Submissions will be added later after the questions are imploded, so that we
+# can explode it without extra duplicates.
+speakers1 = speakers1.drop("submissions", axis=1)
+
+
+speakers2 = implode(
+    speakers1,
+    keep=[
+        "speaker_code",
+        "speaker_name",
+        "email",
+        "bio",
+    ],
+    agg=[
+        "q_id",
+        "q",
+        "answer",
+    ],
+)
+
+print(speakers2)
+
+
+qa = {
+    "company": 2346,
+    "country": 2347,
+    "job": 2348,
+    "homepage": 239,
+    "github": 2352,
+    "linkedin": 2353,
+    "underrep": 2354,
+    "underrep_details": 2355,
+    "underrep_other": 2356,
+    "finaid": 2357,
+    "under18": 2358,
+}
+
+
+for col, key in qa.items():
+    speakers2[col] = speakers2["Q"].map(lambda x: _get_answer(x, key))
+
+speakers2 = speakers2.drop("Q", axis=1)
+speakers2 = speakers2.set_index("speaker_code")
+speakers2["subs"] = temp["submissions"]
+speakers3 = speakers2.explode("subs")
+
+
+# ===========================
+# 2. CLEAN THE QA on SUBS
+# ===========================
+
+with_answers_1 = pd.merge(
+    sub,
+    ans[["q_id", "q", "person", "submission", "answer"]],
+    how='left',
+    left_on="code",
+    right_on="submission",
+)
+
+with_answers_2 = implode(
+    with_answers_1,
+    keep=[
+        "code",
+        "title",
+        "type",
+        "track",
+        "duration",
+        "state",
+        "abstract",
+        "description",
+    ],
+    agg=[
+        "q_id",
+        "q",
+        "answer",
+    ],
+)
+
+
+qa = {
+    "length": 2360,
+    "onsite": 2362,
+    "exp": 2363,
+    "tweet": 2365,
+    "for_reviewers": 2366,
+    "material": 2367,
+}
+
+for col, key in qa.items():
+    with_answers_2[col] = with_answers_2["Q"].map(lambda x: _get_answer(x, key))
+
+with_answers_2 = with_answers_2.drop("Q", axis=1)
+# with_answers_2 = with_answers_2.set_index("code")
+
+
+# ==================================
+# 3. MERGE SUBMISSIONS WITH SPEAKERS
+# ===================================
+
+df = pd.merge(
+    with_answers_2,
+    speakers3,
+    how='left',
+    left_on="code",
+    right_on="subs",
+)
+
+# NOW ON TO THE VOTING AND REVIEWS DATA.
+
+# Reviews is easy - we already have the data, we just need to load and merge it
+# again.
+# For voting we will need to get the export from the voting app and then
+
+
+# =====================
+# 4. MERGE REVIEWS
+# =====================
+
+# Backup the columns, you will need them later for the aggregation.
+cols = df.columns
+
+
+gr = rev.groupby("submission").agg(
+    Rsum=pd.NamedAgg(column="score", aggfunc="sum"),
+    Ravg=pd.NamedAgg(column="score", aggfunc="mean"),
+    Rmed=pd.NamedAgg(column="score", aggfunc="median"),
+    Rcnt=pd.NamedAgg(column="score", aggfunc="count"),
+)
+
+df = df.set_index("subs")
+
+withrevs = df.join(gr, how="left")
+
+
+# ==================================================================
+# 5. groupped scores are done, now we need to show individual scores.
+# For every review we need to create a column
+# ===================================================================
+
+
+reviewers_to_initials = {
+    x: "".join([y[0] for y in x.split()]).upper() for x in sorted(rev["user"].unique())
+}
+cc = Counter(reviewers_to_initials.values())
+
+for k, v in reviewers_to_initials.items():
+    # If there are dupplicates just add numbers
+    if cc[v] > 1:
+        reviewers_to_initials[k] = f"{v}{cc[v]}"
+        cc[v] -= 1
+
+
+rev["initials"] = rev["user"].map(reviewers_to_initials)
+rev.head(2)
+
+reviews_p = rev.pivot(columns="initials", values=["score"], index="submission").fillna("")
+reviews_p.columns = reviews_p.columns.to_flat_index()
+
+main = withrevs.join(reviews_p, how="left")
+
+# =====================================================
+# 6. YAY. ONE LAST FINAL PIECE. COMMUNITY VOTING
+# =====================================================
+
+vot = vot.set_index("code")
+main = main.join(vot, how="left")
+
+
+print(main.columns)
+IDX = [
+    "type",
+    "track",
+    "title",
+    "speaker_name",
+    # "duration",
+    # "state",
+    # "abstract",
+    # "description",
+    "length",
+    "onsite",
+    "exp",
+    "tweet",
+    "for_reviewers",
+    "material",
+    "email",
+    "bio",
+    "company",
+    "country",
+    "job",
+    # "homepage",
+    # "github",
+    # "linkedin",
+    "underrep",
+    "underrep_details",
+    "underrep_other",
+    "finaid",
+    "under18",
+    # Review bulk metrics
+    "Rsum",
+    "Ravg",
+    "Rmed",
+    "Rcnt",
+    # Community voting
+    "score",
+    "must",
+    "want",
+    "maybe",
+    "votes",
+    "comments",
+    # Individual review scores
+    ("score", "A"),
+    ("score", "AC"),
+    ("score", "AH"),
+    ("score", "AM"),
+    ("score", "AWT"),
+    ("score", "CB"),
+    ("score", "CL"),
+    ("score", "CM"),
+    ("score", "CM2"),
+    ("score", "DL"),
+    ("score", "DS"),
+    ("score", "DV"),
+    ("score", "EA"),
+    ("score", "JW"),
+    ("score", "LC"),
+    ("score", "NR"),
+    ("score", "NT"),
+    ("score", "PJ"),
+    ("score", "RD"),
+    ("score", "RD2"),
+    ("score", "RK"),
+    ("score", "RP"),
+    ("score", "SG"),
+    ("score", "SS"),
+    ("score", "SW"),
+    ("score", "SZ"),
+    ("score", "TM"),
+    ("score", "VGF"),
+    ("score", "VM"),
+]
+
+main = main[IDX]
+
+skip_score = {x: x[1] for x in main.columns if x[0] == "score"}
+main = main.rename(columns=skip_score)
+
+# # Clean up remaining few things
+main = main.rename(columns={
+    "length": "_L",
+})
+# This is a question so can't be cleaned up early at the beginning, only after
+# merging with answers and imploding
+main["onsite"] = main["onsite"].map(lambda x: "REMOTE" if "remote" in x else "YES")
+
+# Convert the talk ID to a clickable link that goes to the reviews
+# First we drop the current index (which is a pretalx ID), the new column would
+# be called "index" – we can keep the name, it's good enough.
+# Then we add hyperlinks to this new column before exporting.
+main = main.reset_index()
+
+# Then as a final step apply the link
+def link(x):
+    event_name = "europython-2023"
+    url = f"https://pretalx.com/orga/event/{event_name}/submissions/{x}/reviews/"
+    return f'=HYPERLINK("{url}", "{x}")'
+
+
+main["index"] = main["index"].map(link)
+
+
+# And finally export to an excel file
+tstamp = dt.now().strftime("%Y%m%d-%H%M")
+fname = f"./main-filtered-out-missing-entries-{tstamp}.xlsx"
+main.to_excel(fname, index=False)
+
+print(f"FINISHED AND SAVED at {fname}")
diff --git a/2023/programme/01_generate_spreadsheet_from_pretalx_data/Makefile b/2023/programme/01_generate_spreadsheet_from_pretalx_data/Makefile
new file mode 100644
index 0000000..aa641a8
--- /dev/null
+++ b/2023/programme/01_generate_spreadsheet_from_pretalx_data/Makefile
@@ -0,0 +1,11 @@
+deps/init:
+	pip install pip-tools
+
+deps/compile:
+	pip-compile
+
+deps/install:
+	pip-sync
+
+run:
+	python ./01_download_data_from_pretalx.py
diff --git a/2023/programme/01_generate_spreadsheet_from_pretalx_data/config.py b/2023/programme/01_generate_spreadsheet_from_pretalx_data/config.py
new file mode 100644
index 0000000..ad990b3
--- /dev/null
+++ b/2023/programme/01_generate_spreadsheet_from_pretalx_data/config.py
@@ -0,0 +1,4 @@
+import tomllib
+
+with open("config.toml", "rb") as fd:
+    conf = tomllib.load(fd)
diff --git a/2023/programme/01_generate_spreadsheet_from_pretalx_data/config.toml.example b/2023/programme/01_generate_spreadsheet_from_pretalx_data/config.toml.example
new file mode 100644
index 0000000..a4a2df4
--- /dev/null
+++ b/2023/programme/01_generate_spreadsheet_from_pretalx_data/config.toml.example
@@ -0,0 +1 @@
+pretalx-token = "this is pretalx token - for submissions"
diff --git a/2023/programme/01_generate_spreadsheet_from_pretalx_data/requirements.in b/2023/programme/01_generate_spreadsheet_from_pretalx_data/requirements.in
new file mode 100644
index 0000000..65a42be
--- /dev/null
+++ b/2023/programme/01_generate_spreadsheet_from_pretalx_data/requirements.in
@@ -0,0 +1,2 @@
+requests
+pandas
diff --git a/2023/programme/01_generate_spreadsheet_from_pretalx_data/requirements.txt b/2023/programme/01_generate_spreadsheet_from_pretalx_data/requirements.txt
new file mode 100644
index 0000000..8a8ea6d
--- /dev/null
+++ b/2023/programme/01_generate_spreadsheet_from_pretalx_data/requirements.txt
@@ -0,0 +1,28 @@
+#
+# This file is autogenerated by pip-compile with Python 3.10
+# by the following command:
+#
+#    pip-compile
+#
+certifi==2024.2.2
+    # via requests
+charset-normalizer==3.3.2
+    # via requests
+idna==3.6
+    # via requests
+numpy==1.26.4
+    # via pandas
+pandas==2.2.1
+    # via -r requirements.in
+python-dateutil==2.9.0.post0
+    # via pandas
+pytz==2024.1
+    # via pandas
+requests==2.31.0
+    # via -r requirements.in
+six==1.16.0
+    # via python-dateutil
+tzdata==2024.1
+    # via pandas
+urllib3==2.2.1
+    # via requests